I'm passionate about Machine Learning, Data Science, Large Language Models (LLMs) and Generative AI.

The rapid rate at which there is advancement in computing power and the freedom to take a large volume of data pushes the field of Artificial Intelligence to a whole new level.

I hold a Masters Degree in Data Science at Northeastern University - Khoury College of Computer Sciences.

I have 4+ years of experience building and deploying machine learning and deep learning models.

Furthermore, I have a strong practical and theoretical experience in the development of Large Language Models (LLMs) and Generative AI.

🔭 Below are some of the companies I have worked as a data scientist and a machine learning engineer:

• NVIDIA Corporation
• Solbots Technologies Private Limited
• Khoury College of Computer Sciences

🔭 Some of the notable courses I have completed and that helped in gaining strong theoretical foundation include:

• Machine Learning Certification by Stanford University
• Deep Learning Specialization by Andrew Ng
• Deploying AI & Machine Learning Models for Business from Udemy
• Python for Time Series Data Analysis by Jose Portilla

🔭 I've used different Machine Learning and Deep Learning models in real-time projects. Below are some used models:

• Linear Regression
• Logistic Regression
• Support Vector Machines (SVM)
• Decision Trees (DT)
• Random Forests (RF)
• K-Nearest Neighbors (KNN)
• Deep Neural Networks
• Convolutional Neural Networks (CNN)
• Recurrent Neural Networks (RNN)
• Naive Bayes (NB)
• Gradient Boosted Decision Trees (GBDT)
• XGBoost
• Long Short-Term Memory (LSTM)

🔭 Below are some state-of-the-art (SOTA) time series forecasting models used in various real-time projects:

• Auto-Regressive (AR) Model
• Auto-Regressive Moving Averages (ARMA) Model
• Auto-Regressive Integrated Moving Averages (ARIMA) Model
• Neural Hierarchical Interpolation of Time Series (N-HiTS) Model
• Seasonal Auto-Regressive Integrated Moving Averages (SARIMA) Model

🔭 Furthermore, below are some of the tools used during my experience for Generative AI:

• Langchain
• LangGraph
• Retrieval Augmented Generation (RAG)
• Llama Index
• OpenAI API
• Mixtral (LLM)
• Llama 2 (LLM)
• GPT - 3 (LLM)
• GPT - 3.5 (LLM)
• GPT - 4 (LLM)

🔭 Here are some of the skillsets in regards to DevOps technologies:

• Docker
• Docker-Compose
• Kubernetes
• Linux
• Windows Subsystem for Linux (WSL)
• Travis CI
• Ubuntu
• GitLab CI/CD

These valuable tools and techniques have empowered me to successfully develop and comprehend intricate machine learning projects.

+ The images are downloaded and used thanks to https://unsplash.com/ and https://giphy.com/ websites. 🙂

I encourage you to explore my machine learning and deep learning projects. The links are provided below, along with detailed descriptions at the bottom of this website.

🏥 Health Assistant Application

🚴‍ Washington Bike Demand Prediction	🚙 Car Prices Prediction

🏦 Predicting Loan Default	🫀 Heart Disease Prediction

🏚 Airbnb Home Prices Prediction	☎️ Telco Customer Churn Prediction

📚 Predicting Readability of Texts	👹 Twitter Sentiment Analysis

🗞 Fake News Prediction	🔎 Automated Essay Scoring with Transformers

🎋 Wheat Disease Detection Using Deep and Transfer Learning	🌞 Solar Panels Dust Detection

🌾 Wheat Localization With Convolutional Neural Networks (CNNs)	🥄 Steel Defect Detection

🔢 MNIST Digits Classification	🕸 Convolutional Neural Networks CNN Implementation Using Keras

📚 Article Recommender System

📽 YouTube Video Analysis	🔍 Google Play Store Genre Prediction

🚕 Cab Reservation System

🍿 IMDB Movies Web Scraping	🍔 Restaurant Recipes Web Scraping XML

🎮 Popular Gaming Titles Wikipedia Web Scraping	🎒 University Instructors Information Scraping

🔐 JSON file Web Scraping

🍔 Adare Restaurant Webpage	🏍🚲 Roar Bikes Webpage

   🏢👨‍💻 Data Scientist | NVIDIA [July 2023 - Present]

   👨‍🏫🧑‍🏫 Graduate Teaching Assistant - SML | Northeastern University [September 2022 - December 2022]

   🏢👨‍💻 Data Scientist Intern | NVIDIA [May 2022 - August 2022]

   👨‍🏫🧑‍🏫 Graduate Teaching Assistant - NLP | Khoury College of Computer Sciences [December 2022 - May 2022]

   🧪👨‍🔬 Research Assistant | Khoury College of Computer Sciences [December 2022 - April 2022]

   🤖🦾 Data Scientist | Solbots Technologies [January 2018 - August 2020]

   🏫 Northeastern University (Khoury College of Computer Science) - Masters in Data Science

   🏫 Arizona State University (Ira A. Fulton School of Engineering) - Masters in Software Engineering

   🏫 VNR Vignana Jyothi Institute of Engineering and Technology - B.Tech in Electronics and Communication Engineering

   🏫 Narayana Educational Institutions - High School

   🏫 Vikas The Concept School - Junior High School

   🌱 Machine Learning by Stanford University

   🌱 Deep Learning Specialization

      🍐 Neural Networks and Deep Learning

      🍐 Structuring Machine Learning Projects

      🍐 Sequence Models

      🍐 Improving Deep Neural Networks: Hyperparameter Tuning, Regularization and Optimization

      🍐 Convolutional Neural Networks

   🌱 Python for Data Science and Machine Learning

   🌱 Data Science and Machine Learning Bootcamp with R

   🌱 Machine Learning Engineering for Production (MLOps) Specialization

      🍐 Introduction to Machine Learning in Production

      🍐 Machine Learning Data Lifecycle in Production

      🍐 Machine Learning Modeling Pipelines in Production

      🍐 Deploying Machine Learning Models in Production

   🌱 Complete Tensorflow 2 and Keras Deep Learning Bootcamp

   🌱 Deploying AI & Machine Learning Models for Business | Python

   🌱 Python for Time Series Data Analysis

   🦸 Leadership Skills

   🦸 Communication Skills

   🦸 Team Work

   🦸 Curiosity

   🦸 Problem-solving Skills

   🦸 Time Management

   📃 Machine Learning

   📃 What are the best applications of Machine Learning?

   📃 Why is it just as important for Machine Learning Models to be Fair as well as Accurate?

   📃 Graphical Processing Units (GPUs) can be used for deep learning apart from just gaming

   📃 How important is data in Machine Learning?

   📃 What are Convolutional Neural Networks (CNN)?

   📃 How Machine Learning could be used in the Cloud?

   📃 Best projects to Showcase in Machine Learning Portfolio

   📃 In Machine Learning, Correlation might not always be equal to Causation

   📃 Predicting the Sentiment of a text using Machine Learning

   📃 Why is GPT-3 revolutionizing the Natural Language Processing Industry (NLP)?

   📃 How to understand Machine Learning?

   📃 Introduction to Natural Language Processing for Machine Learning

   📃 Various ways at which Machine Learning could be used in Medical Diagnosis

   📃 What are the activation functions in Machine Learning?

   📃 Predicting the Difficulty of Texts Using Machine Learning and Getting a Visual Representation of Words

   📃 Detect the Defects in Steel with Convolutional Neural Networks (CNNs) and Transfer Learning

   📃 Why is it Important to Constantly Monitor Machine Learning and Deep Learning Models after Production?

   📃 How to Address Data Bias in Machine Learning?

   📃 Understand the Workings of Black-Box models with LIME

   📃 Understanding the Why of Data Science and Machine Learning Is More Useful than Knowing the How

   📃 List of Important Libraries for Machine Learning and Data Science in Python

   📃 Differences between Bias and Variance in Machine Learning

   📃 Common Reasons why Machine Learning Projects Fail

   📃 10 Tips to become a Data Scientist or Machine Learning Engineer

   📃 Do Not Curse Your Machine Learning Models When They are Not Performing Well in Real Time - Instead Do This

   📃 Step-by-step Approach of Building Data Pipelines as a Data Scientist or a Machine Learning Engineer

   📃 Which Feature Engineering Techniques improve Machine Learning Predictions?

   📃 How to Avoid Mistakes in Data Science

   📃 Various Types of Deployment In Machine Learning

   📃 Exploratory Data Analysis (EDA) on Airline Sentiment Tweets

   📃 Unleash the Hidden Patterns: A Guide to Undersupervised Machine Learning for Article Recommender System

   📃 Clearing the Dust: How CNNs and Transfer Learning Can Detect Dust on Solar Panels

   📃 Constructing a Decision Tree Classifier: A Comprehensive Guide to Building Decision Tree Models from Scratch

   📃 Unsupervised Machine Learning: Explore a List of Models that work without Output Labels

   👀👁 Vision Transformers (ViTs)

   📣🔊 LIME & SHAP (Explainable AI)

   📕📚 BERT & RoBERTa

   🔎🔦 FAISS & ScaNN

   🎧🔉 Audio Signal as a Spectrogram

   🤝🕸 Siamese Networks

   📸 My Professional Introduction

   🔗 https://www.kaggle.com/suhasmaddali007

   📄 Data Scientist Resume

   ✅ Bi Senior Foua - Data Scientist at Apple

   ✅ Abhik Lahiri - Machine Learning Lead at PathAI

   ✅ Mano Satya Sai - Chief Executive Officer at Solbots Technologies

   📫 Email: Suhas.maddali.edu@gmail.com

   🖇 LinkedIn: https://www.linkedin.com/in/suhas-maddali/

   📘 Facebook: https://www.facebook.com/suhas.maddali

   ✍🏻 Medium: https://suhas-maddali007.medium.com

   💻 Kaggle: https://www.kaggle.com/suhasmaddali007

   📹 YouTube: https://www.youtube.com/channel/UCymdyoyJBC_i7QVfbrIs-4Q

My interest in machine learning and artificial intelligence began when I was in my final year of a Bachelor of Technology program. I suggested to my teammates that we work on a machine learning project in healthcare, specifically using it to predict the chances of a person suffering from heart disease. We were able to download a dataset from Kaggle and implement machine learning models to make predictions.

The results on the test set were promising, and this sparked my interest in finding more ways to apply machine learning. I have since taken courses in machine learning, including Andrew Ng's Machine Learning and Deep Learning Specialization and have worked on various data science projects in healthcare, academia, and the retail industry. Some of these projects involved data analysis and visualization to gain insights.

📝 Furthermore, I also write articles and share my blogs through Medium. It is a website where writers could share their thoughts with the community through publications. I've written articles on Machine Learning and Data Science. Below is the link to my Medium articles. https://suhas-maddali007.medium.com

In this section, different sets of machine learning projects are highlighted. Feel free to click on the links for the projects that are highlighted. There is a brief description of the projects along with useful definitions. In the next few sections, a subset of Artificial Intelligence such as Computer Vision and Natural Language Processing (NLP) would be covered along with Data Visualization.

🚴‍♂️ Washington Bike Demand Prediction

• In this project, the demand for rental bikes was predicted using different machine learning and deep learning algorithms.
• Moreover, some of the useful features were visualized.
• Dependencies between features were highlighted and a correlation matrix was plotted to get an understanding of the relationship between features.
• Subsequently, machine learning predictions were executed to ensure accurate outputs for the corresponding test set.

🚙 Car Prices Prediction

• In this machine learning project, the prices of cars would be predicted based on features such as Horsepower (HP), Mileage, Make, and other features.
• Most of the project contains visualizations followed by machine learning and deep learning algorithm predictions.
• Based on the prices determined by the algorithms, companies could set the price for the cars which would result in high profitability.

🏦 Predicting Loan Default Using Machine Learning

• It is important for banks to give loans to customers based on their ability to pay back a loan.
• Occasionally, banks encounter situations where they grant loans to individuals who fail to repay the borrowed funds, including the accrued interest.
• Machine Learning could be used in order to determine whether a loan must be given to a person, and this would help the financial institutions and banks to save money respectively.
• In the project, various features such as income levels and the amount of loan taken were considered as features for predicting whether a person would be paying back a loan or not.
• Since the data that was taken contained a lot of NULL values, various imputation methods were used such as mean, median, and mode imputation.

🫀 Heart Disease Prediction Using Machine Learning and Deep Learning

• There are features that are important to predict heart disease in a patient such as Blood Pressure (BP), BMI and other factors.
• Since doctors cannot take into account all the factors and suggest whether a person may or may not have heart disease, it is time to use machine learning and deep learning algorithms to the rescue.
• In this machine learning project, various features such as BMI, Cholesterol, and other factors are considered for predicting the chances of a person suffering from heart disease.
• Various machine learning models were used for the predictions and their precision, recall, and accuracy were plotted respectively.

🏚 Airbnb Home Prices Prediction

• Features such as neighborhood, longitute and latitude could be used to determine the prices of Airbnb houses for hosts.
• Data Science and Machine Learning could be used to extract these insights from data and also make useful predictions for housing prices.
• Since the demand is also an important factor to consider, it becomes really interesting how the prices would vary based on this feature and many others.
• Exploratory Data Analysis (EDA) was performed to take a look at features that make the highest impact when determining the prices of houses along with others.

☎️ Telco Customer Churn Prediction

• Worked on predicting the churn rate of customers based on factors such as age, location, and the type of service that was chosen by the customers.
• Exploratory Data Analysis (EDA) was conducted to gain a comprehensive understanding of all the features and relationships in the dataset. Visualizations were used to effectively communicate insights.
• A large number of metrics were considered such as accuracy, logistic loss, precision, recall, F1 score and many others.
• Vast proportion of machine learning models were used in the process of prediction of churn rate of customers such as Logistic Regression, Decision Tree Classifier, Gaussian Naive Bayes, Random Forest Classifier and XGBoost Regressor.
• Finally, the best model was selected and was hyperparameter tuned to get the best results on the data that the models have not seen before.

Natural Language Processing (NLP) is converting a natural text into a form that could be used for machine learning and deep learning purposes. It involves extracting texts, removing stopwords, lemmatization and stemming, lowercasing the letters, and removing punctuations and other text information that do not add a lot of meaning in our machine learning predictions. Below are the links to some of the Natural Language Processing (NLP) projects.

📚 Predicting Readability of Texts Using Machine Learning

• Text mining, data visualization, and machine learning can reveal useful insights in the vast amount of text that surrounds us, making it more actionable and useful.
• In addition to this, understanding the difficulty of the text and whether it is at our level could give us good knowledge about the depth of the article.
• Moreover, libraries and other educational institutions could use this information and classify textbooks and notebooks, and separate them based on the difficulty level.
• In this project, machine learning and deep learning algorithms were used to predict the difficulty of texts.
• Hyperparameter tuning was also done to ensure that the models took important features into consideration and made predictions with low mean squared error and mean absolute error respectively.

👹 Twitter Sentiment Analysis

• In Twitter, there are comments made for different posts and tweets. Sometimes, there might be negative comments that would change the course of the direction of certain topics.
• It is important to identify comments and extract key features from the text so that positive and negative comments could be separated.
• In this project, we are going to be extracting useful information from texts and understand key components for differentiating between positive texts and negative texts.

🗞 Fake News Prediction

• There has been a lot of fake news that is spread to a large number of people on a regular basis, leading to misinformation.
• With the aid of machine learning and data science, it is possible to extract insights from news.
• Classification ML models are used to determine if the given news is fake or not based on a set of factors.
• Metrics such as accuracy, precision, recall and f1-score are used to analyze different set of algorithms before deploying in production.
• This saves a lot of time for readers as they are not fed with misleading information.

🔎 Automated Essay Scoring with Transformers

• AI analyzes student essays for coherence, clarity, and relevance and offers feedback to improve writing skills.
• Deep learning models (Transformers, LSTMs, GRUs, CNNs, and ML) detect visual cues, model coherence, relevance, and identify factors for effective writing in student essays.
• Multiple approaches provide a comprehensive evaluation, empowering educators to give targeted feedback to improve student writing skills.
• Metrics like accuracy, precision, recall, F1 score, and QWK evaluate model performance for classifying essays.

Computer Vision is a subset of artificial intelligence which gives the computer to perform computations and make predictions on image data. If the image of a cat is given, for instance, the computer vision algorithms would classify whether there is a cat in the image based on a previously labeled set of images. Therefore, it is important to give the right data to the computer vision algorithms in order for them to get the right predictions.

Computer vision offers immense potential and high demand due to its wide range of applications. The abundance of image and video data creates numerous opportunities for leveraging computer vision tools effectively. Here, I present a selection of intriguing projects I have personally undertaken in the field of computer vision.

🎋 Wheat Disease Detection Using Transfer Learning

• Wheat is commonly available in various forms such as cereals and bread, but it is susceptible to diseases during production.
• Manual identification of wheat diseases is time-consuming for farmers.
• Computer Vision can be utilized to understand and identify different diseases in wheat, leading to significant time savings.
• This technology enables prompt identification of diseases and facilitates preventive measures for future occurrences.
• Several Convolutional Neural Networks (CNNs) like InceptionV3, Xception, and VGG19 were tested to detect wheat diseases.
• VGG19 exhibited exceptional performance, achieving a test accuracy of 97 percent.

🌞 Solar Panels Dust Detection

• Solar panels are used in a large number of industries to generate electricity and renewable sources of energy.
• There is a considerable amount of dust and dirt accumulated in the solar panels due to adverse weather conditions, bird droppings, prolonged use, and many other factors.
• With the aid of convolutional neural networks and transfer learning, it is possible to detect dust accumulated in solar panels, leading to efficient maintenance cycles with minimal friction.
• This increases the efficiency of solar panels, giving rise to a higher amount of energy from the solar source.
• Networks such as VGG16, VGG19, MobileNet, AlexNet, Xception and DenseNet are explored to determine whether a solar panel is clean or dusty.

🥄 Steel Defect Detection

• Steel industries produce large quantities of steel, but unnoticed defects can impact quality and lead to issues like corrosion.
• Manual defect detection is time-consuming, necessitating the use of analysis and prediction mechanisms.
• Convolutional Neural Networks (CNNs) can generate predictions based on past training examples.
• Machine learning models, particularly deep learning models, save time and effort by predicting and addressing steel defects.

🌾 Wheat Localization With Convolutional Neural Networks (CNNs)

• There are a lot of food products that we get from wheat.
• Wheat is known as the common food staple that could be used to prepare different kinds of food items.
• Since there are different types of wheat available, it is important to identify different kinds of wheat available.
• In the machine learning project, images of different kinds of wheat heads are taken and made available so that computer vision could be used to understand wheat heads and distinguish them.
• Various computer vision algorithms are used to identify the wheat heads respectively.

🔢 MNIST Digits Classification

• MNIST data is quite popular as it is being used for beginning the journey with Convolutional Neural Networks.
• The repository contains an MNIST project that would classify the images into 9 digits starting from 0 to 9 respectively.
• There are different configurations of Convolutional Neural Networks being implemented by taking into consideration the cross-entropy loss as the metric for getting the best configuration.

🕸 Convolutional Neural Networks CNN Implementation Using Keras

• This is a simple project to implement Convolutional Neural Network and note its working.
• Different layer sizes and different kernels are chosen and trained on a simple dataset. The data that was taken was MNIST which is available in Kaggle.
• There are different kernel sizes considered and outputs are noted using a graph respectively.
• By doing this project, I've learned to use Keras and Tensorflow for building Convolutional Neural Networks (CNNs) respectively.

Unsupervised machine learning is a type of machine learning in which the model is not provided with labeled training examples. Instead, the model is given only unlabeled data and must find patterns and relationships in the data on its own. Unsupervised learning algorithms are used to find patterns in data, group data into clusters, and identify anomalies. Some common unsupervised learning methods include clustering, dimensionality reduction, and anomaly detection.

📚 Article Recommender System

• Built an article recommender system that is able to generate recommendations of articles based on the similarity of previous articles.
• Used methods such as k-means clustering for understanding the right amount of clusters based on elbow method.
• Worked with dimensionality reduction techniques such as PCA and t-SNE.
• Performed exploratory data analysis on the text information to understand the occurrence of words based on similarity.

It is important to note that data visualization is the key to extracting insights from the data. In addition to this, it also gives us insights into whether new features must be created or removed. If we find that there is a strong correlation between different features, all those features can be removed (considering that the dimensionality of the data is very large). Therefore, we have to spend time performing Exploratory Data Analysis (EDA). Below are some of the data visualization projects.

📽 YouTube Video Analysis

• Since there are many videos being uploaded every day, it is important to analyze the videos and the categories.
• In the project, we have also analyzed the likes, comments, and other important features when videos are being uploaded.
• Various plots were used in the process of exploring the data such as Scatterplots and Countplots.

🔍 Google Play Store Genre Prediction

• In this analysis, we are going to understand the different types of apps in the Google Play store.
• In addition, we are also going to understand the NULL values that are present in our data.
• Moreover, we are going to analyze the total number of apps that are paid and free.
• We also would differentiate those apps based on age groups respectively.

It is also important to understand how data is stored before performing machine learning analysis. It could be stored either in Relational or Non-relational format depending on the requirements. It is important to understand the requirements of the customers in order to design a database and perform querying and retrieval of information. You might take a look at some of the ideas expressed below. Thanks.

🚕Cab Reservation System*

• In this project, database design of the cab reservation system was created and analyzed to implement the effective way to store and retrieve information from the database.
• Furthermore, Flask was also used to design and store the data values for storing information about various customers.
• After successfully designing and implementing the database design, the focus was placed on building the UI interface so that users could book cars as and when needed based on the availability of the drivers.

Web scraping is the process of extracting data from websites. This data can be used for various purposes, such as data science and machine learning. There are different methods for extracting data from websites, which may be stored in HTML, XML, or JSON formats. To successfully scrape data from the web, it is important to use the appropriate techniques. Some web scraping projects also involve querying information from the web.

🍿 IMDB Movies Web Scraping

• IMDB rating could be used to analyze user engagement with various movies.
• It would be really good if we could be using the tables from the IMDB data and understand the factors impacting the ratings by various users.
• Web scraping was done with the help of various packages in Python and understanding the data respectively.
• One popular library used to understand the data was BeautifulSoup that was important for reading the tables respectively.

🍔 Restaurant Recipes Web Scraping XML

• There are a lot of items being ordered in restaurants and shopping malls.
• The information is stored on the internet.
• As a result, this could be used for various machine learning and deep learning purposes.
• Web scraping of the XML files was performed in this project.
• In addition to this, it was also important to determine the paths needed for robust querying of the data.
• All of these steps were performed with the aid of packages in Python.

🎮 Popular Gaming Titles Wikipedia Web Scraping

• Wikipedia is a good source of information and is often reliable, especially with the trends.
• Web scraping this information could be handy. Web scraping of Gaming data was performed to extract useful insights from it.
• There were various tables present in Wikipedia. Interest was towards the most trending Games and the overall revenue generated by them.
• It was very interesting to work with the data and understand the games that were in high demand.

🎒 University Instructors Information Scraping

• It would also be very interesting to know the demand for various professors in different departments.
• Scraping the information and understanding the demand along with other features could give us a good idea overall about a professor and their teachings.
• The files which were used for performing the web scraping were XML files and various libraries were used in the process.

🔐 JSON file Web Scraping

• There could be many websites that give APIs to implement their work.
• We could take the APIs and also extract the JSON files that are stored using key-value terminology.
• When this is performed and done, we could smartly extract useful features that are important for the machine learning predictions depending on the project at hand.
• You might take a look at this project as it highlights how to gain useful information from a website with JSON data.

With the help of the tools such as HTML, CSS, and bootstrap, websites are designed to help users with easier navigation. As a result of the design, users end up getting attracted to the interactive UI, thus, booking the services as required. Feel free to take a look at the website design. The back-end programming is not yet done (would be done in the future). An interactive front-end is designed so that users get access to the content.

🍔 Adare Restaurant Webpage

• A website is designed with knowledge of HTML and CSS.
• Styling is also done with the CSS stylesheets that really help in building effective designs.
• Users are able to book a reservation along with locating the exact place where the restaurant is present.
• In addition, a youtube video is also embedded in the homepage so that visitors take a look before ordering the food in the restaurant.

🏍🚲 Roar Bikes Webpage

• Roar bikes is a bike service company that uses online applications to filter the potential candidates for the service of their bikes.
• The front end of the web application is designed along with all the functionalities so that it becomes easy for the user to search and navigate the most appropriate bike service.

👨‍🏫🧑‍🏫 Graduate Teaching Assistant - SML | Northeastern University [September 2022 - December 2022]

• Worked with students in helping them understand the assignments and guiding them in the process to make the concepts of supervised machine learning understandable.
• Regularly attended office hours to discuss with students about their progress in the assignments and also suggesting them changes to be made to get the best results.
• Often had meetings with the professor and other teaching assistants to make modifications to the syllabus and assignments.
• Directed a large number of students in the right direction especially for job applications for internships.

🏢👨‍🏫 Data Scientist Intern | NVIDIA [May 2022 - August 2022]

• As a data scientist at NVIDIA, I was influential in using state-of-the-art machine learning and deep learning models to predict future demand for products based on past history.
• In addition to this, classical models were also used and compared before deciding the best architecture for supply chain optimization.
• Finally, the explainable part of artificial intelligence was implemented with the use of various tools. This ensured that the predictions given by various ML models are made interpretable to the business before they take action from them.

👨‍🏫🧑‍🏫 Graduate Teaching Assistant - NLP | Khoury College of Computer Sciences [December 2021 - May 2022]

• I'm a graduate teaching assistant at Khoury College of Computer Sciences Northeastern for the course Natural Language Processing (NLP CS6120) under professor Uzair Ahmad.
• I'm involved in designing the assignments for students, teaching them concepts, and improving their performance on the tests.
• I teach students classes if a lot of students don't understand the concepts being taught in the classroom.
• I give students the course materials and links that they could use to improve their performance on the exam and the assignments.
• During office hours, I explain the concepts to students and make it easy for them to score well.

🧪👨‍🔬 Research Assistant | Khoury College of Computer Sciences [December 2022 - April 2022]

• I worked as a research assistant for Neural Network (NN) Verification and specification. The research is under Tan Cheng.
• We worked on verifying the execution of neural networks on untrusted servers, black-box databases, and systems.
• I'm involved in understanding the behaviors of Neural Networks in various settings and whether they perform the computations as per the specification.
• In addition to this, I use a programming language called Julia to run the neural network and understand their specifications.
• Overall, it was a good experience to implement neural networks and define their behaviors as well.

🤖🦾 Data Scientist | Solbots Technologies [January 2018 - August 2020]

• I have 2+ years of experience working as a data scientist at Solbots Technologies.
• I was able to influence and direct a team of people in understanding and applying machine learning and deep learning algorithms.
• During my time at the company, I had to develop Convolutional Neural Networks (CNN) to detect images present in front of the Bionic hand.
• The input from the camera would contain a list of images and videos of different objects.
• Later, the deep learning model would make its predictions, and accordingly, the bionic hand would adjust the grip.
• Furthermore, I had to gain a theoretical and practical understanding of machine learning and deep learning before implementing them in the company.
• It was a good experience working in the company and I gained a good amount of knowledge in the field of data science and machine learning.

🏫 Northeastern University (Khoury College of Computer Science) - Masters in Data Science

I have pursued a Master's in Data Science at Khoury College of Computer sciences. During the course, I've learned to implement various machine learning algorithms. Furthermore, the courses that were taught really shaped the way in which I approached machine learning and data science problems. I had an opportunity to learn machine learning algorithms from a theoretical point of view by doing assignments and projects pertaining to the course. It also helped me learn some important courses that were really influential in my progress toward learning the concepts of data science. Courses such as Supervised Machine Learning, Unsupervised Machine Learning, and Data Visualization helped me understand the practical ways in which data science is used in industrial applications.

🏫 Arizona State University (Ira A. Fulton School of Engineering) - Masters in Software Engineering

I pursued a Master's in Software Engineering at Arizona State University. It was a good experience where I could get a solid understanding of working with software engineering principles and paradigms. Furthermore, I was also involved in a project where we as a team had to build an AR application in real time. It was a good experience where I learned software engineering principles and learned to work in sprints for a different set of tasks respectively. Furthermore, I also completed the data structures and algorithms course which allowed me to understand the time complexity of various algorithms which was also influential in my understanding of machine learning algorithms respectively.

🏫 VNR Vignana Jyothi Institute of Engineering and Technology - B.Tech in Electronics and Communication Engineering

I've completed a Bachelor of Technology in the field of Electronics and Communication Engineering(ECE) and got a good understanding of the concepts such as resistors and conductors and how they influence the overall circuit in real-life. During my 4 year journey in Electronics and Communication Engineering, I came to the conclusion that we would be learning various communication devices such as satellites and circuits that would influence the ways in which a mobile signal is being received for mobile applications. Furthermore, this gave me a good understanding of signal processing and other important topics that could also be used in machine learning analysis respectively.

🏫 Narayana Educational Institutions - High School

During high school, I had a good opportunity to join an institute that focused mostly on mathematics, physics, and chemistry. I was able to learn those concepts really well and get a good understanding of the theory behind the working of various concepts. It was a highly focused course where students had to get good grades in order to continue their education at the institute. I was able to score well on the Mains exam - an exam that is highly competitive in India. It was in part due to my hard work and the teachings by the institute to prepare students to excel well in the exam. Overall, this really shaped the way in which I approach mathematics problems and get a solid understanding of the concepts respectively.

🏫 Vikas The Concept School - Junior High School

During my education at Vikas The Concept School, I had a good foundation in mathematics, physics, social sciences, and Indian languages. Furthermore, we also had extracurricular activities that really shaped the way in which I viewed education, and they made a good impact. I was also involved in student leadership activities that really gave me good knowledge about how to handle people and derive outcomes in efficient ways. I was also involved in dance clubs and singing clubs where I gave a live performance in front of a group of people. This gave me a lot of confidence and I was able to somehow manage the fear of public speaking. All in all, my school was one of the best experiences where I could learn theoretical concepts and see how they are being used in real life.

There are numerous machine learning and data science courses that I went through in order to gain a theoretical understanding of the concepts before their practical implementation in the form of projects. Below are some of my certifications and the contents covered in the course respectively.

🌱 Python for Data Science and Machine Learning - This course is taught by Jose Portilla. It is a course that gave me a very good understanding of Python. Important topics such as data frames, tuples, and lists were discussed in the course. The instructor does a good job of showing the practical implementation of the course along with theory. Therefore, this gave me a good solid understanding of Python which later helped me to build machine learning and deep learning algorithms.

🌱 Deep Learning Specialization - This is a really long certification consisting of 5 courses. The courses give a good understanding of deep learning. The courses are taught by Andrew Ng. Furthermore, there were videos that were being uploaded about interviews with pioneers in the field of Deep Learning. General tips and advice for machine learning projects are also laid out to ensure that beginners don't make the mistake of performing tasks that may not be required. Overall, I was able to get a good understanding of all the courses and they gave me a good idea about Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Deep Neural Networks, and Long-Short Term Memory (LSTM) networks. Below is the list of 5 courses that are present in the specialization.

   🍐 Neural Networks and Deep Learning - In the first course of the specialization, Andrew Ng concentrates on Logistic Regression, Shallow Neural Networks, and Deep Neural Networks for performing various computations and getting the results. He first explains the basics of how a neuron performs complex computations which would, in turn, lead to predictions. This course gave me a good understanding of the theory behind the working of activation units in neural networks.

   🍐 Structuring Machine Learning Projects - It is important to take measures to improve the performance of the deep learning models. In this second course under Deep Learning Specialization, structuring machine learning projects and understanding bias, variance, and practical applications of deep learning are taught by the instructor. Overall, it gave me a good foundation to apply my machine learning knowledge to practical real-time projects.

   🍐 Sequence Models - In the 3rd course of the specialization, Recurrent Neural Networks (RNNs), Word Embeddings, and Natural Language Processing Techniques are taught by the instructor. Moreover, he gives good clarity of different embedding techniques before giving the data to the deep learning models for predictions. Sequence Models and Attention Mechanism topics are also covered in the course. This gave me a good conceptual and practical understanding of deep learning and data science.

   🍐 Improving Deep Neural Networks: Hyperparameter Tuning, Regularization and Optimization - When performing the deep learning computations, it is always a good idea to perform hyperparameter tuning in order to get the best results for our test set. In the course, the importance of regularization and hyperparameter tuning is taught by the instructor. Furthermore, various optimization algorithms are taught by the instructor. All-in-all, it was a good course that covered the important aspects of deep learning in a data science life cycle respectively.

   🍐 Convolutional Neural Networks - In the final course of the Deep Learning Specialization, the instructor teaches the foundations of Convolutional Neural Networks (CNNs). Also, Deep Convolutional Neural Networks along with case studies are included in the lectures. Finally, Face Recognition Technology and Neural Style Transfer are taught in the course, giving a good idea of their work. Hence, I was able to get a good amount of working knowledge in the field of data science and deep learning by going through all the courses and completing the specialization.

🌱 Machine Learning by Stanford University - This is a course taught by Andrew Ng. I was able to understand the theory behind machine learning and deep learning models. Furthermore, he also gives practical advice on how machine learning could be used in different industries. The language that was used for programming was Octave. Overall, this gave me a good understanding of machine learning and I was able to enter the field starting with this course.

🌱 Data Science and Machine Learning Bootcamp with R - R is a programming language that could be used for statistical purposes. I was able to implement the machine learning models using R. There are various ways in which R programming language is used in different scenarios. The was taught by Jose Portilla. It gave a good insight into using R for machine learning purposes. Most of the videos are focused on the practical implementation of the machine learning models respectively.

🌱 Machine Learning Engineering for Production (MLOps) Specialization - The specialization opens new frontiers in machine learning by teaching set of steps and methodologies in building highly scalable and robust end-to-end machine learning systems. A lot of new topics were covered in the space of MLOps such as model decay, latency, system requirements and many others. The framework used for end-to-end deployment was tensorflow extended (TFX), leading to efficient and highly scalable solutions to various ML problems.

   🍐 Introduction to Machine Learning in Production - This course gives examples of how to use various strategies for building machine learning models in production. There are often challenges that go unnoticed if care is not taken when models are put to production. As a result, there is a degradation in their performance. This course does a good job in highlighting this scenario in great detail along with steps to be taken to reduce them.

   🍐 Machine Learning Data Lifecycle in Production - The course introduces various components in the TensorFlow Extended (TFX) library that is used by end-to-end deployment. Components such as StatisticsGen, SchemaGen, ExampleValidator, and ExamplesGen are used which form the initial pipeline components of TensorFlow Extended. A lot of exercises were given to get a firmer understanding of these components in great detail in many scenarios.

   🍐 Machine Learning Modeling Pipelines in Production - This course teaches the best practices to follow when building data pipelines for machine learning. Depending on the size and complexity of the network, various set of practices must be followed in order for the models to work well in specific platforms.

   🍐 Deploying Machine Learning Models in Production - This course primarily deals with building and deploying machine learning models in production. Knowledge about various inference engines and serving mechanisms were explained and demonstrated well. Kubernetes and docker were used as important tools for ML deployment along with providing the right security features for different network topologies.

🌱 Complete Tensorflow 2 and Keras Deep Learning Bootcamp - Deep learning has been gaining traction in the recent decade. I could see that most of the projects on Kaggle are done with Tensorflow and Pytorch. The course is taught by Jose Portilla from Udemy. Most of the course is focused on the implementation of deep learning using Keras and Tensorflow. I got a good understanding of implementing various deep-learning projects using Keras and Tensorflow.

🌱 Deploying AI & Machine Learning Models for Business | Python - This course teaches the fundamentals of Docker and explains them in detail. In addition, the instructor does a good job in explaining how Docker along with Flask and Apache could be used for deploying our machine learning models and making them highly scalable. It also talks about various errors that can occur during the production environment that is important to consider when trying to deploy machine learning models to a large number of clients and users.

🌱 Python for Time Series Data Analysis - This course emphasizes the construction of time series models using up-to-date data from diverse sources. Participants gain a comprehensive understanding of cutting-edge models, including ARMA, ARIMA, SARIMA, SARIMAX, and Deep Neural Networks, for accurate forecasting. The course is highly regarded, offering clear explanations of the concepts and delivering valuable insights.

🦸 Leadership Skills

• During my B.tech in India, I had an opportunity to direct a team of students in our final project proposal.
• It was a good opportunity for me to improve my leadership skills during the process.
• I improved my knowledge in the field of machine learning and also learned the skills needed to direct people so that we get the best results respectively.

🦸 Communication Skills

• Communication skills are also important to become a good software engineer and a data scientist.
• Communication plays a very important role when it comes to letting others in the team know the progress, tracking the development of a project, and getting a good understanding of the overall flow of the team.
• I built my communication skills during my masters where I had to discuss my assignments and projects and let them know the overall scenario where our project could be used.
• Furthermore, I've gone through courses that are related to communication which would ensure that we are getting the best results when talking to a group of people or an audience respectively.

🦸 Team Work

• When building projects and talking to people about the outcomes, it is important to have teamwork so that it would be a whole lot easier for a team to improve the performance of the company.
• That's the reason why companies such as Apple and Facebook are improving their revenue as a result of work from the team rather than individual efforts.
• Given my strong involvement in teams, I have actively participated in numerous team-based projects, honing my networking and team management skills along the way.

🦸 Curiosity

• Being creative when building applications would lead to better and more innovative products.
• Some of the most remarkable breakthroughs take place with curiosity in the field.
• I believe that having high levels of curiosity in endeavors could lead to better outcomes not only in the short run but in the long run as well.

🦸 Problem-solving Skills

• These are the skills that generally help in solving issues quickly and effectively.
• They are being learned as part of education or training.
• They are generally about familiarizing themselves with the common issues in various industries and also learning from experienced employees.

🦸 Time Management

• During my tenure at Solbots Technologies, I earned a professional reputation by learning to manage time and giving the best outcomes for the company by working smart and trying to get things done in less time.
• Therefore, I was involved in the process of organizing and planning various activities.
• As a result, there was very good productivity in the company by managing time and effectively using it for best practices.

There is a lot of resources and tools available in the world with the advancement of technology. I believe that education should be accessible to everyone regardless of their location, age and social status. It is possible to discover very good insights with the help of machine learning and data science and use them to serve education in different parts of the world. I also believe that each and every individual is unique and outstanding in his/her ways. Each and every individual must be respected regardless of their conditions or their significance in society. All in all, I believe that one must give respect to each other and this would ensure that we go in the right direction and make a significant impact in society.

My interest in machine learning started during my final year of engineering at Vignana Jyothi Institute of Technology. During that time, we had to apply different machine learning and deep learning techniques for the prediction of heart disease in patients. I've also found interest in writing blogs and articles and sharing my knowledge with the community so that newbies in machine learning could read and understand them. Below is a list of blogs and articles that I've written along with a short description.

📃 Machine Learning

• In this article, basic definitions of machine learning and deep learning are mentioned.
• Overfitting and Underfitting are some of the challenges in machine learning.
• Their definitions and various ways in which those could be avoided are also talked about in the article.
• There are different types of machine learning such as supervised machine learning, unsupervised machine learning, and semi-supervised machine learning respectively.
• Their definitions and usefulness are mentioned.

📃 What are the best applications of Machine Learning?

• There has been a significant increase in the number of machine learning applications in the recent decade.
• In the medium article, I highlight some of the potential applications of machine learning.
• Machine learning is being used in diverse fields such as social media, image detection, and traffic prediction.
• All these topics are mentioned in the article.

📃 Why is it just as important for Machine Learning Models to be Fair as well as Accurate?

• There are machine learning and deep learning models that might have good accuracy but might be biased.
• It is important to address the data that is being provided to the machine learning models for prediction.
• It is also useful to give data that depicts real-world use cases respectively.
• Sometimes the data that is given to the machine learning models might be biased.
• Therefore, steps must be taken to ensure that models take data that is unbiased.
• I've highlighted various ways in which machine learning models might be biased. Feel free to click on the link to get started.

📃 Graphical Processing Units (GPUs) can be used for deep learning apart from just gaming

• GPUs are quite often used to run graphics-intensive games and used by gamers.
• However, it is important to note that they could also be used to perform deep learning computations.
• In the article, I've mentioned some of the differences between a CPU and a GPU.
• In addition, I've described how the computations would increase with the help of GPUs as compared to that CPUs.

📃 How important is data in Machine Learning?

• Data is present all around us. Companies generate a ton of data in different ways which could be used for machine learning and deep learning purposes respectively.
• In the medium article, the importance of data is shown.
• Moreover, there could be data in different forms that could be used for machine learning.
• Data could be either in categorical forms or numerical format.
• Various steps must be taken for various data points in order to get the outputs.
• Various types of data are described.

📃 What are Convolutional Neural Networks (CNN)?

• Computer Vision is applying deep learning models in order to classify objects in images and videos.
• It is important to understand Convolutional Neural Networks (CNNs) so that we could use them for our machine learning tasks.
• In this article, Convolutional Neural Networks are described.

📃 How Machine Learning could be used in the Cloud?

• There are a lot of cloud-based services that are present.
• We see companies such as Amazon and Google releasing their cloud services where the infrastructure is provided by them.
• In the medium article, I’ve highlighted how the cloud could be used for machine learning.
• Since the field of data science requires a good amount of computational resources, it would be wise to use cloud-based platforms to perform training and test machine learning and deep learning models.

📃 In Machine Learning, Correlation might not always be equal to Causation

• When taking into account visualizations from different features in our data, we could sometimes assume that correlated features are dependent on each other and occurrence of one feature might lead to the occurrence of the other.
• However, this might not always be true when performing the visualizations.
• In the blog, I’ve written examples and illustrations where correlation might not always be equal to causation and explained the details.

📃 Predicting the Sentiment of a text using Machine Learning

• There is an abundance of text around us. Since the text is in the string format when talking in terms of computer context, it is important that steps are taken to convert it into a form that could be used for machine learning purposes.
• We use various natural language processing techniques for the process of converting the raw text into a form that is easily feasible for machine learning algorithms.
• In the medium article, I’ve mentioned some of the latest natural language processing techniques.

📃 Why is GPT-3 revolutionizing the Natural Language Processing Industry (NLP)?

• Text is available in many forms such as in books, articles, magazines, and publications.
• It would be really wise to make use of the text that is present in industries.
• We've seen how Facebook has released the GPT algorithm that would take the text and understand the context, making its predictions in the future.
• In the medium article, I've mentioned the use of GPT by giving numerous examples of its working respectively.

📃 How to understand Machine Learning?

• In this article, I discuss various courses that teach machine learning such as courses by Andrew Ng and Jose Portilla.
• Moreover, various steps are mentioned about the proper use of these courses to gain a good understanding of machine learning algorithms.
• It is important to learn the right courses before diving into machine learning projects.
• Having a good theoretical knowledge of machine learning models and optimization techniques ensures that one gets a firm understanding of the practical implementation of models.

📃 Best projects to Showcase in Machine Learning Portfolio

• This article discusses the various types of projects that could be done by people who want to enter the field of data science and machine learning.
• In addition, it also explains the various projects in a clear fashion so that people who are new to machine learning could establish their grip in the process of implementing the models respectively.

📃 Introduction to Natural Language Processing for Machine Learning

• There is an abundance of text present in magazines, newspapers, articles, and blogs.
• In order to use the natural text, conversion mechanisms must be followed which would ensure that we get the best results when performing machine learning analysis.
• In the article, I've mentioned various natural language processing steps and ways to process raw text and convert it to a form that could be used for machine learning purposes.

📃 Various ways at which Machine Learning could be used in Medical Diagnosis

• Machine learning and data science could be used in different domains.
• It is important to note that machine learning could be used in medical diagnosis where patients could be treated by taking a look at various features that are important for machine learning predictions.
• In the medium article, I've highlighted how machine learning and data science could be used in the medical industry along with their potential challenges.

📃 What are the activation functions in Machine Learning?

• We see that there are a lot of applications in machine learning.
• There is complex math involved in the field of artificial intelligence.
• As a result, it would be highly valuable if activation functions are understood.
• These are mostly used in Neural Networks and also machine learning.
• In the medium article, I've highlighted various activation functions along with their equations.

📃 Predicting the Difficulty of Texts Using Machine Learning and Getting a Visual Representation of Words

• Sentences and paragraphs are present in all of our textbooks and notebooks.
• It would be handy if their difficulty is understood so that the most appropriate text could be given to children depending on this feature.
• However, reading and understanding the difficulty of texts manually is time-consuming.
• Therefore, we would be using machine learning and data science for predicting the difficulty of texts.
• In the article, I've highlighted how we could be using NLP and machine learning for text classification based on the difficulty.

📃 Detect the Defects in Steel with Convolutional Neural Networks (CNNs) and Transfer Learning

• Steel is available almost everywhere from buildings, roads, and appliances.
• During the manufacture of steel, there are defects that go unnoticed.
• As a consequence, this leads to lower-quality steel that is less durable and also affected by environmental conditions.
• With machine learning and convolutional neural networks (CNNs), it is possible to detect the defects in steel which results in saving a lot of time and money.

📃 Why is it Important to Constantly Monitor Machine Learning and Deep Learning Models after Production?

• While it can be impressive to get the results from ML models for various set of business use cases, failing to put the models in production is almost equivalent to not using ML at all.
• In this blog, I highlight the importance of monitoring the performance of machine learning models after production and ensuring that the results are not quite different from the models that are initially trained.

📃 How to Address Data Bias in Machine Learning?

• Most of the ethical questions that are raised when using machine learning and data science can be answered with the help of understanding the datasets that were initially used to train the models.
• Understanding and acknowledging that the dataset is biased can help reduce a large portion of the questions for ethical AI.
• This blog highlights the ways in which datasets can be manipulated or modified so as to reduce the inherent bias present in them.
• Therefore, we are assured of getting ML models that are representative of the whole population.

📃 Understand the Workings of Black-Box models with LIME

• Sometimes the machine learning models are able to give very strong predictions that are comparable to the human level performance.
• However, there can be instances where people who are involved might want to know the reason why the models gave a particular set of predictions.
• One of the concerns with machine learning models is that they are not explainable or interpretable though they have impressive performance on the test set or the data that the models have not seen before.
• With the help of LIME, we can interpret the results of any ML model from simple to the most complex state-of-the-art models. This blog explains the coding implementation of LIME used for explainable AI.

📃 Understanding the Why of Data Science and Machine Learning Is More Useful than Knowing the How

• Knowing the purpose of machine learning is useful before we can use these tools for our business or our company. Before we start applying machine learning to a large number of applications, it can be useful to know why it is used in the first place.
• Understanding this can help transition from a good data scientist to a great data scientist.
• This article mainly discusses the importance of this question and gives various examples for the readers.

📃 List of Important Libraries for Machine Learning and Data Science in Python

• There are a huge number of libraries present for machine learning. Understanding the most useful from them is handy and makes our work more productive.
• In this article, I highlight a list of all the libraries that are often used in Python specifically for machine learning and data science respectively.

📃 Differences between Bias and Variance in Machine Learning

• In this article, the differences between bias and variance are discussed in great detail along with the steps that could be taken to reduce them.
• These are the most common issues faced when trying to build machine learning and deep learning models.
• Therefore, this article explains these concepts in great detail so that practitioners can take the right steps to reduce them to a large extent.

📃 Common Reasons why Machine Learning Projects Fail

• In order to build and deploy machine learning models, there are certain challenges that must be addressed so that we are hassle-free when trying to build them.
• Addressing these challenges can be the primary goal of an ML practitioner.
• In this article, we deal with some of the common ways in which machine learning projects fail.
• Though it is impressive to get the models to get the best predictions, not taking care of a few details can have an impact in the long run.
• Furthermore, this article also lists various steps that can be taken to reduce the occurrence of failure.

📃 10 Tips to become a Data Scientist or Machine Learning Engineer

• In this blog, I have discussed some of the important ways in which we can become data scientists or machine learning engineers.
• There are some tips that must be followed that can increase the chances of becoming a data scientist to a large extent.
• While learning to code can be an important skill, the ability to communicate well with team members and other people can make a significant impact on becoming a data scientist respectively.

📃 Do Not Curse Your Machine Learning Models When They are Not Performing Well in Real Time - Instead Do This

• Oftentimes there is a possibility that the machine learning models that were performing well on the training data do not perform well on the data that they have not seen before in real-time.
• This is the case where there are other effects taking place such as data drift and concept drift respectively.
• This article highlights key pieces of information about how to deal with such scenarios in great detail respectively.

📃 Step-by-step Approach of Building Data Pipelines as a Data Scientist or a Machine Learning Engineer

• There is a list of steps that are usually followed by data scientists and machine learning engineers to build interesting artificial intelligence applications.
• In the article highlighted, a sequence of steps that are usually followed in the data science cycle is mentioned to make it easier to implement machine learning solutions.

📃 Which Feature Engineering Techniques improve Machine Learning Predictions?

• Feature engineering can be one of the most important steps in a machine learning cycle.
• Learning to use the right strategies for feature engineering boosts the performance of these intricate models to a large extent.
• In the article, a list of a large number of feature engineering techniques is presented to the community so that they could be used later when needed.
• Following these steps can make it easier to develop a machine learning cycle and get better predictions from a large set of ML models.

📃 How to Avoid Mistakes in Data Science

• Data Science and Machine Learning are used in a large number of industries.
• Practitioners of machine learning and data science sometimes can make mistakes along the way when they are building interesting artificial intelligence applications.
• In this article, I explain a list of mistakes that can occur and some practical tips to avoid them.

📃 Various Types of Deployment In Machine Learning

• This article mainly focuses on building highly scalable machine learning models using various deployment strategies.
• We also take a look at some of the key considerations that are important for deployment in various conditions.
• For example, deploying an ML model in the cloud could be quite different as compared to deploying the models with mobiles or edge devices.
• Therefore, tools and techniques of deployment should be changed depending on where they should be deployed.

📃 Exploratory Data Analysis (EDA) on Airline Sentiment Tweets

• This focuses on exploring the data from airline tweets to understand the sentiment of customers towards various flight companies.
• It explores the overall data and the amount of positive, negative, and neutral tweets.
• It takes a look at the different categories of flights and their percentage out of all the flight categories.

📃 Unleash the Hidden Patterns: A Guide to Undersupervised Machine Learning for Article Recommender System

• Artificial intelligence and machine learning are advancing rapidly and have great potential for generating value in various fields.
• Unsupervised machine learning, including recommendation systems, is a less talked about but important aspect of AI.
• This article focuses on building an article recommender system using unsupervised machine learning techniques like cosine similarity scores between articles.

📃 Clearing the Dust: How CNNs and Transfer Learning Can Detect Dust on Solar Panels

• Dust on solar panels reduces efficiency, but automation and deep learning can detect and alert authorities for timely maintenance.
• Libraries like TensorFlow, NumPy, Pandas, and OS simplify development and improve the effectiveness of models for solar panel dust detection.
• These tools enable building complex and accurate models with greater ease and efficiency, leading to more effective renewable energy generation.

📃 Constructing a Decision Tree Classifier: A Comprehensive Guide to Building Decision Tree Models from Scratch

• In this article, I delve into the critical role that decision tree classifiers and regressors play in data analysis and machine learning.
• To ensure a thorough understanding of decision tree models, this article not only explains their inner workings but also provides code examples for readers to follow along.
• Through practical coding exercises, readers will gain a solid grasp of the fundamental concepts that underpin decision tree models, making it easier to apply them in their own projects.

📃 Unsupervised Machine Learning: Explore a List of Models that work without Output Labels

• We explore a list of unsupervised machine learning models in this article. In addition, we also understand their working in detail.
• Pros and cons of each and every algorithm are explored in order to determine the right model based on business contraints.
• Models such as k-means, gaussian mixture models, hierarchical clustering, and others are deeply covered.

During my machine learning journey, I had a good time learning important things and takeaways while implementing and executing various projects. As a result, I was able to learn iteratively and update my knowledge of the latest technologies and tools used in the process of building interesting AI-powered applications. Given below are some of the repositories that I have added that I felt had key ingredients in them that helped me excel in this data science journey.

📽 Jose Portilla's Reinforcement Learning Course - One of the interesting things about the instructor Jose Portilla is his attention to detail and clarity of explanation. Reinforcement learning has a lot of potential, especially in database systems and computer systems. After going through the course, I learned a lot of intricate details about how to define an agent and an environment which are the key tools in reinforcement learning. You might take a look at the repository where I present the notebooks which were used for learning the basics and advanced concepts related to python.

There is a large volume of research taking place in the field of machine learning and data science. There are newer and computationally efficient algorithms being developed by the likes of many companies and research institutes. I would like to share my thoughts on these latest machine learning trends and explain them well.

   👀👁 Vision Transformers (ViTs) - Transformers have revolutionized the natural language processing industry (NLP) where a given text is converted into a representation that takes into account the contextual information for all the possible words given as input and returns a vector with these weights and other dependencies. One interesting research area that has emerged is to use of these same transformers for computer vision tasks. Convolutional Neural Networks, CNNs for short, are currently being used to take into account different positions of the image and map them with their weights before making predictions. But if we could represent these vectors by using contextual dependencies, then vision transformers might be able to replace CNNs in the future. Currently, as the performance of vision transformers has not been significantly higher than the CNN models, there is no replacement for CNNs. However, as the complexity of vision transformers increases, there is a possibility that they might replace CNNs for image processing tasks.

   📣🔊 LIME & SHAP (Explainable AI) - LIME stands for Local Interpretable Model-agnostic Explanations while SHAP stands for Shapley Additive Explanations. One of the challenges with using AI and machine learning, in general, is their lack of interpretability. Though there are models such as Random Forests and Decision trees that explain why they have come up with a particular outcome such as giving feature importance, they fail to account for local dependencies which means they do not give output for a particular query point but only provide explainability in terms of the entire dataset. Furthermore, there are other models that do not also offer these features. With the help of LIME and SHAP, it is possible to explain for a query point why a particular outcome is generated for any of the machine learning models at hand. Therefore, LIME and SHAP are good, and executing them is also a lot easier with the help of libraries.

   📕📚 BERT & RoBERTa - BERT stands for Bidirectional Encoder Representations from Transformers while RoBERTa stands for Robustly Optimized BERT Pre-training Approach. These models have been gaining popularity due to their extremely good performance in natural language processing tasks. They are basically transformers with bidirectional context vector representation. In other words, they take into account the context in terms of both the forward pass and the backward pass as well. This works extremely well due to the fact that representing each word based on the context of all the words in the document is basically a good way to understand human language.

   🔎🔦 FAISS & ScaNN - When we are dealing with finding similar rows in our data and we have a very high dimensional representation of a vector, with traditional algorithms it takes a long time to find similarly if we use metrics such as euclidean distance and cosine similarity. Facebook Research has launched FAISS that simplifies the process of searching for similarity and clustering of dense vectors. In the same light, Google AI Research came up with ScaNN which speeds up the overall process of computing the distance between various features and reduces the time complexity to a large extent. The outputs from these models contain vectors that represent a large amount of information from the input data. These are useful for building good recommender systems where the items that are the most similar are recommended to a user along with many other applications as well.

   🎧🔉 Audio Signal as a Spectrogram - We can define our audio signal in terms of a spectrogram which is used in audio processing. One of the most insightful things that are currently done for deep learning is that an audio signal is converted to a spectrogram image which is later used with convolutional neural networks (CNNs) for NLP tasks. Therefore, we are trying to pose the audio problem as a computer vision problem and get higher accuracy and better results. CNNs are known to perform especially well if we are able to give large amounts of image data. CNN along with transfer learning produces extremely good results. Therefore, it is improving the performance of NLP tasks such as speech detection and many others.

   🤝🕸 Siamese Networks - Building recommender systems whether it be recommending items, movies, or songs with the help of deep neural networks can be a hard problem to define and solve respectively. With the help of Siamese neural networks, it is possible to use deep learning for recommending users various items as discussed above. Siamese networks essentially take into account 2 different inputs and they have 2 networks that accept these inputs. The same weights are initialized to both networks and the outputs from these two networks are combined together to get our predictions of how likely is the user going to like a particular item. Furthermore, it is also possible to perform one-shot learning with these models which means that with few training examples, the model would learn to perform the best for the task at hand.

Machine learning is used in many industries and there is a lot of demand and scope for it. There are so many tools and resources that one could be used in order to become a machine learning engineer. However, there are certain challenges in machine learning that must be addressed before they could be used for machine learning analysis.

1. Availability of data - Getting access to quality data is important for machine learning predictions. Sometimes, companies cannot have access to a large volume of data that could be used for machine learning purposes. Moreover, the data that is available might not give a very good prediction accuracy and might not be quite useful for the machine learning predictions at hand.

2. Infrastructure Requirements - There is also a possibility for the companies particularly startups to have the infrastructure needed for the machine learning processes respectively. They might be used some third-party applications such as Amazon Web Services (AWS) in order for them to perform their day-to-day predictions using ML algorithms. However, this would lead to further questions such as security and data ownership.

3. Rigid Business models - Companies need to be flexible where they must spend their time with other resources apart from the fixed number of resources for them to work effectively. There are many companies that might require deep learning models to be more flexible and more useful and robust. When we are implementing the machine learning models, there could be instances where we might have to fix our approach rather than be more flexible. This creates problems for large-scale businesses as they must adapt to the market as and when needed. Therefore, steps must be taken to make the models more flexible and easy to understand.

4. Lack of Talent - Machine Learning Engineers and Data Scientists are some of the talked about professions in the IT industry. There are many universities and colleges making use of artificial intelligence programs so that they could meet the demand from the tech industry. However, there still seem to be a lot of gaps in the industry. More and more courses and online learning resources should be created so that more people can enter the field of data science and machine learning.

In the repository, I've created my introduction video so that employers could take a look at it and get to know my core strengths, passion and key knowledge in the field of data science and machine learning in general. Feel free to download the file and take a look at the video. Below is the link for my professional introduction repository. Feel free to download the files and view the mp4 files. Thanks.

https://github.com/suhasmaddali/Professional-Introduction-Repository

I'm very familiar with Kaggle - a website that would help data scientists and machine learning engineers to explore the data and perform machine learning predictions. I would be taking my time to read the kaggle projects which give me a good idea about the working of machine learning models. Furthermore, I would read others' code as a result of which I get a good understanding of the various ways in which a machine learning project could be implemented. Below is a list of all the work that I did in Kaggle. Feel free to take a look and give your feedback! Thanks.

https://www.kaggle.com/suhasmaddali007

👨‍💼 VNR SF

• I was a member of VNR SF and I was most influential in driving the goals of the team in helping the community by providing the needy access to resources and books.
• This gave me a good experience of working with different individuals and learning about their goals and dreams.
• There were a few floods in our region during those times. We as a team went to a few neighborhoods and gave them access to new books and also explained our initiative to serve the poor and needy by giving them access to all the resources.
• This experience was really useful for me and the way in which I viewed education in shaping my life.
• This gave me an opportunity to take a look at the poorest neighborhood and understand their needs and demands.

🎤 Crescendo

• In our engineering college, we had Crescendo which was the singing club where people from different departments in college came together to sing and learn to enjoy music.
• I was the lead singer during that time.
• Furthermore, I was also directing different groups of people to sing and also play musical instruments respectively.
• It was a really good experience to learn from the team the mistakes and correct them and ensure that everything was going well.
• It shaped the way in which I viewed music which was to not just sing but also enjoy the process.
• My 2 years of working with Crescendo really helped in handling practical aspects of life as well.
• Overall, it was a good and amazing experience to learn from a group of people from different departments in my college and it was really influential.

During my work and my LinkedIn contributions, I came across a wonderful set of people who allowed me to exhibit my talents and skills and made my learning experience easy. Furthermore, they were also influential in giving me the right guidance and direction which shaped my interest in machine learning and data science. Below are some wonderful people who were influential and learned about my work in the field of data science and artificial intelligence.

✅ Bi Senior Foua - He is a Data Scientist at Apple and saw my work through LinkedIn Contributions. He gave me a good idea about connecting with people through LinkedIn and Networking events. Furthermore, I discussed my projects with him and got a good idea about some of the improvements that could be made in real-time projects. This gave me a good idea about the overall workflow of machine learning and data science starting from Visualization to the Deployment of the models.

✅ Abhik Lahiri - He is currently working in PathAI as Senior Machine Learning Scientist And Technical Lead and he saw my work through LinkedIn contributions and sharing of blogs in the field of machine learning and artificial intelligence. Furthermore, he also recognized my work through Kaggle and GitHub where I demonstrated my projects and experience in using machine learning and deep learning algorithms respectively. Overall, I had a good time interacting with him and learned a lot during the process. To add more information about Abhik, he also worked in Quora as a Machine Learning Engineer and pursued a Master's in Computer Science at Stanford University.

✅ Mano Satya Sai - He is the Chief Executive Officer (CEO) of Solbots Technologies. During my tenure at Solbots Technologies, I had a very good opportunity to speak with Mano about my knowledge of machine learning and how it could be used in Bionic Hands to detect the type of objects that were present in front of the image or video respectively. He analyzed my work and gave me a team where I could work in machine learning so that we could drive the best outcomes for the company.

Furthermore, I'm also a gamer and where I spend some time gaming on steam and other platforms. Sometimes it would be good to spend time doing activities other than reading which would ensure that we could most effectively use our time for doing other activities as well. Below are some platforms where I usually game mostly.

https://www.ibm.com/cloud/learn/machine-learning

https://www.salesforce.com/eu/blog/2020/06/real-world-examples-of-machine-learning.html

https://www.springboard.com/library/machine-learning-engineering/how-to-become/

https://machinelearningmastery.com/what-is-deep-learning/

https://healthinformatics.uic.edu/blog/machine-learning-in-healthcare/

https://www.analyticsvidhya.com/blog/2021/05/convolutional-neural-networks-cnn/

These are some of my projects, blogs and certifications that I have worked on and uploaded on GitHub. I would be looking forward to learning new technologies in the field of AI and machine learning by going through a few more courses and applying my knowledge to different projects. Feel free to reach out if you have any questions or need any explanations of the projects. Looking forward to sharing my knowledge with the community.

Below are some of the ways we might connect. Feel free to share your thoughts. Thanks!😁

🖇 LinkedIn: https://www.linkedin.com/in/suhas-maddali/

📹 YouTube: https://www.youtube.com/channel/UCymdyoyJBC_i7QVfbrIs-4Q

📫 Email: Suhas.maddali.chinnu@gmail.com

📘 Facebook: https://www.facebook.com/suhas.maddali

✍🏻 Medium: https://suhas-maddali007.medium.com

😄😄😄😄😄😄😄😄😄😄😄😄