I build AI-powered systems and full-stack products around a simple engineering question: can a messy real-world workflow be turned into a system that is faster, clearer, and more useful?

That has led me from semantic search across 15,000 research papers to edge-aware computer-vision intelligence, academic analytics, agentic workflows, and AI tools for learning.

• 🛡️ Hackathon: Team Lead — #14 out of 6,000+ global participants at HackerRank Infinity Hacks 2026
• 🎓 Education: B.Tech CSE, Maharaja Agrasen Institute of Technology
• ⚙️ Current obsession: AI systems, retrieval pipelines, backend architecture and shipping things that actually work
• 🤝 Open to: ambitious builds, hackathons, AI/ML engineering and full-stack collaboration

AI-assisted tactical intelligence fusion for fragmented operational data.

Built around edge vision, SIGINT retrieval and grounded LLM summaries. The system combines YOLOv8, FAISS, FastAPI and Next.js into a single situational-awareness workflow. #14 / 6,000+ globally.

An academic analytics platform designed to make university results usable.

Provider-based university integrations, synchronization infrastructure and a scalable architecture built with Next.js, TypeScript, Prisma, PostgreSQL and Auth.js.

Semantic search and research intelligence over 15,000 machine-learning papers.

A cached FAISS pipeline reduced system boot time from ~30 minutes to under 3 seconds, with BART summarization and KeyBERT extraction layered on top.

Document intelligence for students.

Transforms raw study material into structured notes, active-recall flashcards, quizzes and revision sheets using FastAPI, Next.js and Gemini-powered workflows.

I like projects where the technical problem is only half the problem. The other half is figuring out whether the system survives real users, bad inputs, ugly constraints and the inevitable “wait, what happens if...?”.

Currently learning deeper DSA, system design and production AI engineering while continuing to build.