Full-Stack Engineer • AI Enthusiast • Systems Programmer
Turning ideas into production-grade systems — from kernel-level tools and real-time OpenGL engines to AI-powered platforms serving real users in agriculture and finance.
- Design and ship modern full-stack applications with Next.js 15, TypeScript, and scalable backends.
- Solve real-world problems using Machine Learning and Generative AI (XGBoost, PyTorch, Google Gemini).
- Create beautiful, interactive experiences with OpenGL, particle systems, and advanced AI behaviors (A*, Dijkstra).
| Domain | Project | Highlights | Tech Stack |
|---|---|---|---|
| AI for Agriculture | AgriVision AI | Crop yield prediction, pest & disease detection, soil analysis, and market intelligence for Indian farmers | Next.js 15, Google Gemini, TypeScript |
| Graphics & Game AI | Pac-Man (C++ + OpenGL) | Fully featured Pac-Man with smart ghost AI, particle effects, power pellets, and original soundtrack | C++17, OpenGL, CMake |
| Developer Analytics | DashCode | Competitive programming dashboard with LeetCode/Codeforces tracking, streaks, and global leaderboards | Next.js, TypeScript, PostgreSQL |
| FinTech Security | UPI Fraud Detection | Real-time fraud detection model (87% F1-score) on Indian UPI transactions | Python, XGBoost, Scikit-learn, Flask |
| Campus Collaboration | UniSlot | Real-time slot booking + chat app with AI content moderation for university students | Node.js, Firebase, JavaScript |
I am committed to writing clean, maintainable, and efficient code that solves real-world problems.
- Adaptive quantization techniques in TinyML: Researching dynamic, per-layer, and runtime-aware quantization strategies to push state-of-the-art accuracy on microcontrollers (< 512 KB RAM) while reducing energy consumption by 3–5× compared to static 8-bit methods.
- Post-mortem and live memory forensics: Developing lightweight tooling and algorithms for data recovery, anomaly detection, and reverse engineering in constrained embedded environments.
