📊 Data Every Day

A daily practice series where I work on different datasets, apply machine learning techniques, and keep improving my skills in data preprocessing, modeling, and evaluation.

Along with weekday data practice, every Saturday & Sunday I focus on fine-tuning large language models (LLMs) to strengthen my skills in modern NLP and transfer learning.

The main goal is to stay consistent, learn by doing, and build a portfolio that shows real progress.

---

🚀 Objectives

• Practice data analysis, cleaning, and visualization every day.
• Experiment with ML/DL models on diverse datasets.
• Perform LLM fine-tuning on weekends (Sat–Sun).
• Build a collection of reproducible notebooks for interviews and learning.
• Develop the habit of hands-on problem solving.

---

📂 Repository Structure

Each folder/day includes:

• Dataset / Source
• Exploratory Data Analysis (EDA)
• Data Cleaning steps (sometimes quick or minimal)
• Modeling & Evaluation
• Key Learnings

Weekend Folders (Sat–Sun):

• Fine-tuning experiments with models like BERT, Qwen, GPT-style models, etc.
• Training scripts & configs
• Evaluation reports

---

⚠️ Note

This repo is about learning and consistency.
Some notebooks may have:

• Rough or rushed cleaning steps.
• Quick experiments instead of polished code.
• Areas marked for future improvement.

The focus is on showing progress, not perfection.

---

🧑‍💻 Skills Covered

• Data preprocessing (missing values, scaling, encoding)
• ML algorithms (Logistic Regression, Random Forests, SVMs, Neural Nets, etc.)
• Deep learning basics (CNNs, RNNs, Transformers)
• LLM fine-tuning (LoRA, QLoRA, PEFT, TRL)
• Evaluation metrics (Accuracy, Precision, Recall, F1, ROC-AUC, BLEU, Perplexity)
• Iterative experimentation (Kaggle-style approach)

---

📈 Why It Matters

• Shows commitment to daily practice
• Covers a wide variety of datasets and problems
• Adds LLM fine-tuning experience for cutting-edge NLP skills
• Demonstrates growth over time
• Useful for interview discussions on approach and learnings

---

🌟 Future Plans

• Clean up and refine selected projects.
• Add automated evaluation helpers.
• Expand fine-tuning to more advanced LLMs.
• Share selected notebooks as Kaggle kernels or blog posts.

---

🔗 This repo represents my journey of learning data every day and fine-tuning models every weekend, moving closer to becoming an AI/ML scientist.