Hello, I'm Mahesh!
M.S. Computer Science, UW–Madison, 2026
I recently completed my M.S. in Computer Science at the University of Wisconsin-Madison. I am currently a Research Scientist Intern at Trata (YC W25), where I work on post-training and evaluation of language-model agents in long-horizon financial environments. In September, I will join the MATS OpenAI Stream.
My research focuses on reinforcement learning and post-training for language models, RL generalization, LLM evaluation, and AI safety.
News
- Sep 2026 Joining the MATS OpenAI Stream as a Research Fellow.
- Aug 2026 Started as a Research Scientist Intern at Trata (YC W25), working on agentic financial benchmarks and model post-training.
- Jun 2026 Playing with Fire: What Transfers When RL Trains a Language Agent? accepted to RLxF: Reinforcement Learning from World Feedback at ICML 2026.
- Apr 2026 Sparks of Cooperative Reasoning accepted at ICML 2026!
- Feb 2026 Started as ML Intern at Kalpa Labs (YC F25) in San Francisco.
- Dec 2025 Sparks of Cooperative Reasoning accepted at NeurIPS 2025 LAW Workshop.
- Oct 2025 CuRe accepted at ICCV 2025.
- Jun 2025 CuRe accepted at CVPR DemoDiv Workshop (Oral).
- Feb 2024 MABViT accepted at Deployable AI Workshop, AAAI 2024 (Oral).
Publications
-
Playing with Fire: What Transfers When RL Trains a Language Agent?ICML 2026 Workshop · RLxF: Reinforcement Learning from World Feedback
-
Sparks of Cooperative Reasoning: LLMs as Strategic Hanabi AgentsICML 2026
-
CuRe: Cultural Gaps in the Long Tail of Text-to-Image ModelsICCV 2025 · CVPR DemoDiv Workshop (Oral)
-
MABViT: Modified Attention Block Enhances Vision TransformersDeployable AI Workshop, AAAI 2024 (Oral)
Education
-
University of Wisconsin-Madison Sep 2024 – May 2026M.S. Computer Science
-
Indian Institute of Technology Madras Jul 2019 – Jul 2023B.Tech Chemical Engineering
Experience
-
MATS OpenAI Stream Sep 2026 – Dec 2026Research Fellow
-
Trata (YC W25) Aug 2026 – Sep 2026Research Scientist InternAgentic financial benchmarks and model post-training.
-
Kalpa Labs (YC F25) Feb 2026 – May 2026Machine Learning Intern, San FranciscoLarge-scale generalized audio model training; dense and MoE architectures.
-
JP Morgan Chase & Co Aug 2023 – Aug 2024Quantitative Research Analyst, Asset & Wealth Management
-
WorldQuant LLC May 2022 – Jul 2022Quantitative Research Intern, Mumbai
-
Dassault Aviation, IIT Madras Jan 2022 – May 2023B.Tech Thesis
Competitions & Red Teaming
- OpenAI GPTOSS-20B Red Teaming Hackathon: Honorable Mention (Top 20 / 611)
- OpenAI Bio Bug Bounty: Participant in biological risk red-teaming program
- Frontier Lab Red Teaming: Invited participant in an additional frontier model red-teaming program