Skip to content
View Aarya-Sutar's full-sized avatar

Highlights

  • Pro

Block or report Aarya-Sutar

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Aarya-Sutar/README.md

Aarya Sutar

B.Tech student at the National Institute of Technology Tiruchirappalli, learning AI, machine learning, and full-stack development. I work with Python, Java, Flask, and basic front-end tools, and I’m currently focusing on building chatbots and RAG-style question-answering systems as learning projects. I’ve worked with PDF and ticket-style data ingestion, vector-store retrieval, and API-based backends at an internship, and I’m continuing to improve my skills by building small but practical projects and understanding how they work end-to-end.

Quick facts

  • NIT Tiruchirappalli, Metallurgical & Materials Engineering.
  • ex-AI/ML Intern @ Birlasoft — built a retrieval-augmented Oracle EBS support chatbot with PDF-based document ingestion and workflow diagram automation using draw.io.
  • Primary: Python, Java, Flask, FastAPI, Docker, SQL, prompt engineering, vector-store retrieval, embeddings, REST APIs.
  • Email: sutaraarya5@gmail.com.
  • LinkedIn: linkedin.com/in/aarya-sutar.
  • GitHub username: Aarya-Sutar.

What I do

I am a B.Tech student learning AI and machine learning while building small, practical systems that actually work. I know front end development, Flask, Python, Java and OOP, and I focus on applying that knowledge to chatbots and simple full stack tools rather than abstract research.

  • Currently learning and practicing: retrieval augmented generation, prompt engineering, document and ticket ingestion, and deploying model-backed APIs.
  • Building: the Aurenia chatbot (RAG pipeline, PDF and ticket ingestion, relevance filtering) and a few front-end and Flask projects to sharpen full stack skills.
  • How I work: prototype fast, iterate with tests and simple deployments, document what I build so others can run it. I prefer solving concrete problems over perfect theory.
  • Looking for: internships and real projects where I can learn production practices, improve engineering discipline, and ship features that users can actually use.

If your project needs someone who can bridge ML models and production engineering, I deliver results.

Featured projects

Aurenia Chatbot — PDF-grounded Conversational Assistant (Flask + Ollama)

Repo: Aurenia-Chatbot (or aurenia-chatbot)

What it does

  • Uploads PDFs and extracts text
  • Chunks document content with overlap for context retention
  • Retrieves relevant chunks using heuristic scoring (token overlap, phrase + fuzzy match)
  • Restricts LLM responses strictly to retrieved PDF content
  • Detects weak or missing evidence and blocks ungrounded answers
  • Falls back to general knowledge only after user confirmation

Technical highlights

  • Flask backend with session-isolated documents
  • Chunk-level retrieval with custom scoring logic
  • Intent classification for PDF vs general queries
  • Evidence verification to enforce answer grounding
  • Local LLM inference via Ollama (llama3:8b)
  • Minimal frontend chat UI with safe markdown rendering and PDF upload
  • No external vector database or cloud services

Why it matters: Demonstrates how to build a controlled, source-aware LLM application that prioritizes correctness over fluent hallucinations when working with private documents.

Flask-Blog

Repo: Flask-Blog

  • Full-stack Flask app with authentication, CRUD, pagination, profile management, and secure password reset via email
  • Good example of clean project structure, basic deployment steps, and README-driven onboarding

Other small projects

  • Simon-Game — interactive JS memory game, clean front-end demo
  • ToDo — simple task manager showing UI + persistence basics
  • Dice-Challenge — JS game demonstrating DOM manipulation and game logic

Tech stack & tools

  • Languages: Python, Java, JavaScript, SQL
  • Modeling & ML: prompt engineering, embeddings, vector stores, transformers, fine-tuning basics
  • Web & APIs: Flask, FastAPI, REST, OAuth basics
  • Infra & deployment: Docker, CI automation, simple cloud deployment patterns
  • Data: PDF parsing, CSV/Excel ingestion, metadata extraction, ticket system integration

Running the Project Locally

Prerequisites

  • Python 3.9+
  • Ollama installed and running

Setup

git clone https://github.com/Aarya-Sutar/aurenia-chatbot.git
cd aurenia-chatbot

python -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install -r requirements.txt

Environment Variables

ollama serve
python app.py

Open http://127.0.0.1:5000 in your browser.

Each repo has a full README with endpoints, expected env vars, and demo data.

What I am working on

  • Hardening the Aurenia Chatbot for multi-tenant usage and better access controls
  • Automating ingestion pipelines to handle noisy enterprise PDFs and email/ticket exports
  • Learning production ML ops patterns: monitoring, model versioning, and cost-aware serving

How to collaborate

  • Star or fork the repos you find useful
  • Open issues for bugs or feature requests — I respond faster to concrete, reproducible reports
  • Want me to contribute? Send a short PR or an issue explaining the goal and I will review it

If you want to work together

I take internships and freelance work related to AI/ML systems and frontend/backend engineering. If you need production-ready AI tooling and clean engineering, email me at sutaraarya5@gmail.com or message on LinkedIn.

Pinned Loading

  1. veritas veritas Public

    Python 1

  2. Flask-Blog Flask-Blog Public

    A full-stack Flask blog application implementing user authentication, authorization, CRUD operations, pagination, profile management, and secure password reset via email. Built with Flask, SQLAlche…

    HTML 2

  3. Simon-Game Simon-Game Public

    This is a classic memory-pattern game inspired by “Simon Says.” At each level, the game shows you a sequence of colored blocks. Your job is to repeat the exact pattern by clicking the blocks in the…

    JavaScript 1

  4. ToDo ToDo Public

    todo list, where you can add, delete, strikethrough your daily tasks. Total tasks, and completed tasks are also visible

    CSS 1