Nimey Khurana

At Delhi Technological University, I’m researching Graph Neural Networks and Graph Transformers for link prediction in heterogeneous graphs. I’m investigating GCN, GraphSAGE, GAT, and attention-based architectures to model structural and semantic information.

I built Venu, a full-stack event ticketing and venue booking platform that prevents double-booking with Redis-based temporary seat holds, PostgreSQL transactions, row-level locking, and database constraints. I also implemented real-time seat-state updates with Socket.IO and server-side Razorpay signature verification.

I built TubeMind, a YouTube Retrieval-Augmented Generation system, with transcript ingestion, FAISS vector search, and CrossEncoder reranking. Its FastAPI backend uses SQLite for chat history, with LangSmith for tracing and evaluation.