At ChessPe, I built a computer vision pipeline that reads handwritten chess score sheets, checks move legality, and converts games to PGN. I also architected the platform, including its NestJS and PostgreSQL backend, real-time updates, and coaching reports.
For ChessPe AI, I trained six PyTorch models on an NVIDIA H200 using Stockfish-labelled Lichess games. The tactics model reached 84.1% top-1 accuracy, and I deployed the models with ONNX Runtime for CPU inference.
I also built SiteJaanch, an AI website intelligence platform, and developed a financial-news sentiment classifier using NLP and pretrained LLMs; that research was published in IJARIIT. I'm pursuing an M.Tech in Artificial Intelligence at Bennett University and hold AWS Certified AI Practitioner and AWS Certified Cloud Practitioner certifications.

