Full-stack Developer · Applied AI · Builder · Portfolio · About me . Resume
B.Tech CSE (AI-ML) at VIT Bhopal, graduating July 2026. I build full-stack web apps and ML systems — sometimes separately, increasingly together.
I shipped Hisaab Pro — an offline-first double-entry accounting desktop app — to real clients. I built Skillence with a custom PyTorch model and then replaced the PyTorch serving layer with pure NumPy in production to cut backend crashes to near-zero. I write a lot of tests — Hisaab Pro has 475 Jest tests, not because someone told me to, but because client financial data has no room for silent bugs.
- Offline-first desktop apps — Electron, SQLite WAL, AES-256 encryption, atomic writes
- Full-stack web products — Next.js, React, Node.js, FastAPI, TailwindCSS
- ML systems for production — PyTorch for training, custom NumPy inference for serving
- AI-integrated apps — RAG pipelines, Gemini API, Claude, pgvector, Langfuse tracing
Hisaab Pro — offline-first double-entry accounting desktop app. AES-256 encrypted SQLite, WAL mode, 475 Jest tests, shipped to real clients across 2 product versions.
Skillence — career platform processing 30K+ job postings. Custom PyTorch model, pure NumPy inference layer (~1ms latency), ML salary predictor across 25+ countries, 692-skill × 894-occupation mapping.
SwapHub — campus marketplace for students to buy, sell, donate, and rent items. Express REST APIs, MongoDB, Passport.js auth.
This portfolio — built with Next.js, embedded RAG chatbot using Gemini 1.5 Flash, Supabase pgvector, hybrid search, Langfuse tracing. All free tier.
Why I Wrote 475 Tests for a Desktop Accounting App — when the books are wrong, the app doesn't crash. It just lies to you.
Open to freelance projects and internships or Jobs. I respond same day.