Swarnadeep Saha

Swarnadeep Saha

Researcher at FAIR at Meta. Interested in Large Language Models, Reasoning, Planning

About Me

I am a Research Scientist on the Alignment team at FAIR, Meta. I work with Dr. Jason Weston on Reasoning, Memory, and Alignment of Large Language Models.

I recently obtained a PhD in Computer Science from the University of North Carolina at Chapel Hill, advised by Prof. Mohit Bansal. My PhD was supported by a Google PhD Fellowship and a Rebecca and Munroe Cobey Fellowship.

I study Machine Learning and NLP. Broadly, I am interested in developing language agents that can perform complex, multi-step reasoning and planning tasks (often referred to as the System 2 Reasoning). I see human language as the ideal medium for reasoning, communication, and collaboration among agents. Below are some of the main topics that I’ve worked on (publications here):

  1. LLM-as-a-Judge and Generative Reward Models
  2. Reasoning and Planning in Multi-Agent systems
  3. Attributable Generative Reasoning
  4. Structured Reasoning over Implicit Knowledge
  5. Deductive Reasoning

Recent News