Full Stack Developer · RAG & AI Systems · Hardware guy
Building real-world apps at the intersection of web, AI, and hardware — from India 🇮🇳
- 💻 Full Stack Developer from India
- 🤖 Currently getting deep into RAG pipelines, AI-powered apps & AI systems
- ⚡ I love building real-world web, desktop & hardware-integrated projects
- 🛠 AI Systems · Web Apps · CLI tools · Embedded Systems
- 🚀 Leveling up with Next.js + TypeScript + LangChain
- ☕ Debugging with coffee & lo-fi beats
Ask questions. Get answers. Strictly from your own Markdown files.
A production-ready Retrieval-Augmented Generation (RAG) API and chatbot that turns your local .md knowledge base into an intelligent Q&A engine.
Tech: Python · FastAPI · LangChain · ChromaDB · OpenAI / Gemini / HuggingFace
- 🔌 Pluggable LLM Providers — Switch between OpenAI GPT-4o-mini, Google Gemini 2.5 Flash, or fully local open-source models
- 🧩 Intelligent Chunking —
RecursiveCharacterTextSplitterwith 800-char chunks and 150-char overlap - 🎯 MMR Search + LLM Reranking — Fetches top 12 chunks via Maximum Marginal Relevance, reranks to top 5 with a custom LLM prompt
- 🔁 Idempotent Ingestion — MD5 hashing prevents duplicate vector store entries
- 📉 Context Compression — Eliminates redundant context to reduce token cost
- 📄 Auto Swagger docs at
/docs, CORS-ready, async FastAPI backend
Control computer games with Ultrasonic Sensor + ESP32 + WebSockets
- Real-time hardware ↔ browser communication
- HTML, CSS, JS frontend + C++ firmware
- WebSocket-based live interaction
Modern Next.js + TypeScript batch image editor
- Optimized image processing pipeline
- Clean modern UI with multi-file handling
- Scalable architecture
Python + Tkinter Question Paper Generator
- Create, edit & manage MCQs
- Local database storage
- Desktop UI application