Final-year IT undergrad who likes building things that ship — backend systems with real concurrency guarantees, AI pipelines with real eval numbers, and a macOS app that got picked up by the press.
- 🎓 Final-year IT undergrad @ VIT Vellore (graduating 2026)
- 🔨 Building backend systems and AI-powered applications — FastAPI, LangGraph, RAG, multi-agent pipelines
- 🚀 Shipped CircleSearch, a macOS AI screen-analysis app featured by French Mac publication VVMac
- 🌱 Into distributed systems, vector databases, concurrency, and LLM evaluation
- 🔀 Contributor to open source — PRs to
supabase-py(2.5k★) andfastapi-limiter(750★) - 🪖 NCC Lance Corporal — NCC "C" Certificate holder
| Project | Description | Stack |
|---|---|---|
| CircleSearch | macOS app — drag any screen region for instant AI analysis. On-device OCR, BYOK, multi-provider. Featured by VVMac. | Swift · SwiftUI · Apple Vision |
| Dino Wallet | Double-entry ledger service — concurrency-safe under load (k6: 50 VUs, 0% errors, zero deadlocks) | Python · FastAPI · PostgreSQL |
| Mega AI | Four-agent LLM pipeline with a 15-case eval harness (0.749 vs 0.388 baseline) and a self-improving meta-agent | FastAPI · Celery · Redis · pgvector |
| NOVA | Production RAG over 3,800+ chunks — hybrid BM25 + vector, RRF fusion, Cohere rerank (RAGAS faithfulness 0.89) | FastAPI · pgvector · Next.js |
