I make models faster and lighter on memory. Right now that means draft-and-judge LLM decoding for a 30B diffusion MoE at Conscious Engines, 1.35× end-to-end. Before that, mixture-of-experts neural operators at IBM Research. Dual degree in Electrical Engineering at IIT Bombay.

• COSMOS: first author, NeurIPS 2026 workshop (STODY). Soft mechanism mixtures for long-horizon PDE forecasting.
• Seeing but not speaking: accepted at WACV 2027.

• Amazon ML Challenge · 5th on the public leaderboard Matched 1.7M businesses across three noisy sources: retrieval, LoRA-tuned Qwen judges, LightGBM.
• Gemini Hackathon · Ghar, Live API track Talk to a live AI designer, get room concepts, narrated tours and an architect's build pack. Demo
• Amazon HackOn S6 · ReVival Seven agents that grade, price and resell returned products instead of sending them to landfill.
• ContextCon YC Hackathon · Vantage, Techfest IIT Bombay Ask one question about a company or market and get an answer with receipts. Live

• Took a hotel operating system from zero to one and shipped it to live properties.

anupamrawat.com · X · LinkedIn · anupam.rawat.iitb@gmail.com