Varun Gohil

varunprofilepic2.png

Hi! I am a Ph.D. student at MIT CSAIL working with Prof. Christina Delimitrou. I design and optimize datacenter infrastructure by leveraging machine learning to drive system performance and efficiency.

My current research explores:

  • Resilient ML-driven systems: Improving reliability of systems that use ML policies for autonomous decision-making.
  • LLM-driven heuristic design: Using LLMs to discover and synthesize novel heuristics for system management.
  • Accelerator design and programming: Exploring LLM-driven accelerator design and defining programming abstractions that simplify hardware optimization.

I earned my Master’s at MIT, where I developed a frequency-aware request scheduler for microservices by modifying the Thrift RPC framework. I completed my undergraduate studies at IIT Gandhinagar, during which I worked on approximate computing, memory systems, and hardware for autonomous vehicles with Profs. Manu Awasthi, Joycee Mekie and Rajeev Balasubramonian

News

Jan 26, 2026 Our paper on building resilient ML-driven sytems by using model uncertainty has been accepted to MLSys 2026.
Jan 15, 2026 I’ll be a TA for 6.S984: Datacenter Computing at MIT in Spring 2026.
Aug 29, 2025 Presented our work on improving the reliability of ML for Systems at the MIT/Jane St Research Symposium.