About

I’m a Master’s student in Artificial Intelligence at Northeastern University, and most of my work circles one question: how do you know a result is real. That applies whether the thing being measured is a model I trained or an agent about to write to someone’s database.

I’ve built each layer at least once: a 394M-parameter language model pretrained from scratch (corpus, tokenizer, architecture, training loop), fine-tuning and quantization, retrieval, agents with tool use, constraint solvers, and the FastAPI, Docker, AWS, and Next.js work needed to put any of it in front of users.

That 394M model, sqlpup, now has an official entry on the BIRD leaderboard, and AgentRelBench, my work on agent reliability, was accepted as a poster at a NeurIPS 2026 workshop. In both, the headline finding went against what I was hoping for. Right now I’m the Founding AI Engineer Intern at Interlock Systems (Harvard Innovation Labs), building agent automation and verification over customers’ production ERP systems. Before Boston, I earned my CS degree at BITS Pilani’s Dubai campus.