An answer without a source is just a belief — I build systems that show their work.
I build production-grade AI/ML systems and full-stack applications — not tutorials, not demos. 3rd-year B.Tech CSE (Data Science) @ NSUT Delhi, targeting SWE roles at top-tier tech and finance firms.
- AI/ML in production — built an agentic RAG system with hybrid retrieval + LLM reranking, an in-browser data analyst agent with 5-provider LLM failover (196 passing tests), and a from-scratch ResNet-18 with live Grad-CAM running entirely client-side
- Real-world deployment — HeatWatch India (urban heat ML system) submitted to ISRO Bharatiya Antariksh Hackathon 2026, with satellite data pipelines validated against ground truth
- Engineering discipline over demo-ware — every major project ships with automated test suites, CI pipelines, and honest documentation of trade-offs and known limitations
- 235+ DSA problems solved in C++ (across 3 platforms), actively building toward SDE interview readiness
If it's in this profile, it's deployed, tested, and I can walk you through every design decision behind it.
Languages
Full-Stack
AI / ML
Databases, Cloud & Tools
| Project |
Stack |
Links |
Agentic RAG Assistant

Full-stack retrieval-augmented generation system with hybrid retrieval, LLM reranking, query rewriting, and self-verification — every answer traces back to its exact source.
|
Node.js, React, TypeScript, Groq, Pinecone, Supabase |
|
AI Data Analyst Agent

In-browser data analysis agent that writes and self-corrects SQL/Python against your data, with multi-provider LLM failover and zero server-side execution.
|
React, TypeScript, DuckDB-WASM, Pyodide |
|
Vision Interpretability Studio

Zero-cost, in-browser tool for seeing inside a vision model — live Grad-CAM and feature visualization on a ResNet-18 trained from scratch, running entirely client-side.
|
React, TypeScript, ONNX Runtime Web, PyTorch |
|
Urban Heat Mitigation

AI/ML system analyzing urban heat islands across Indian cities and optimizing cooling strategies. Built for ISRO BAH 2026.
|
Python, XGBoost, React |
|
Cryptex File Sharing
Secure, token-based file sharing platform built with the MERN stack and Supabase Storage.
|
Node.js, Express, MongoDB, Supabase |
|
Context Engineering Toolkit

Full context-engineering pipeline — selection, compression, summarization, and evaluation — built on a genuinely $0 infrastructure stack.
|
React, Node.js, Express, Tailwind CSS |
|
View more projects on my portfolio