resume / github / email scholar / twitter

Aradhye Agarwal

I am a Research Fellow at Microsoft Research, India. I am a part of the reasoning team, under Akshay Nambi, where we work on cutting-edge research in reasoning models and AI agents.

Before this I worked with Prof. Tanmoy Chakraborty, who was my thesis advisor, and the work done by us was published at TACL'24.

Earlier, I had the opportunity to work with Prof. Chetan Arora, with whom I published my very first paper in CVPR'24 on monocular depth estimation.

News

May 1, 2026	My first-authored safety paper is accepted at ICML'26.
Nov 26, 2025	I am invited to the xAI hackathon in SF, CA!
July 21, 2025	I joined Microsoft Research as an RF under Akshay Nambi.
July 17, 2025	Our PEFT paper is officially accepted at TACL'24.
June 26, 2025	I successfully defended my thesis at IIT Delhi!
Dec 20, 2024	I came second in a kaggle competition winning $10,000!
Nov 12, 2024	This webpage is live.

Publications

(* denotes equal contribution)

Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use
Aradhye Agarwal, Gurdit Siyan, Yash Pandya, Joykirat Singh, Akshay Nambi, Ahmed Awadallah
International Conference on Machine Learning (ICML), 2026
paper | arxiv | website | cite

ECoDepth: Effective Conditioning of Diffusion Models for Monocular Depth Estimation
Suraj Patni*, Aradhye Agarwal*, Chetan Arora
Computer Vision and Pattern Recognition (CVPR), 2024
paper | arxiv | code | website | cite

The Third Monocular Depth Estimation Challenge
Computer Vision and Pattern Recognition Workshop (CVPRW), 2024
paper | arxiv | website | cite

Step-by-Step Unmasking for Parameter-Efficient Fine-tuning of Large Language Models
Aradhye Agarwal*, Suhas Kamasetty Ramesh*, Ayan Sengupta*, Tanmoy Chakraborty
Transactions of the Association for Computational Linguistics (TACL), 2024
paper | arxiv | code | cite