Anush Kini

profile_pic.jpg

I am currently collaborating with Prof. Ilija Bogunovic at University College London on distributionally robust optimization in preference learning. Prior to this, I was a Research Fellow (Pre-Doctoral researcher) at Microsoft Research, India working with Dr. Nagarajan Natarajan and Dr. Sayak Ray Chowdhury working on problems in robustness and black-box optimization. Prior to my role at Microsoft, I worked at Google on the Data Commons project, contributing to making public datasets more accessible and usable. I hold a B.Tech in Computer Science and Engineering from PES University, where I graduated with honors.

News

May 2, 2024 Provably Robust DPO: Aligning Language Models with Noisy Feedback has been accepted at ICML 2024! 🥳🎉
Mar 4, 2024 We will be presenting Provably Robust DPO: Aligning Language Models with Noisy Feedback at the ME-FoMo workshop at ICLR 2024.
Dec 1, 2023 OPPerTune is now open-source!. Check out the repository here.
Oct 31, 2023 Our work GAR-meets-RAG Paradigm for Zero-Shot Information Retrieval is up on arXiv.

Publications

  1. Sayak Ray Chowdhury*, Anush Kini*, and Nagarajan Natarajan
    2024
  2. Gagan Somashekar, Karan Tandon, Anush Kini, and 6 more authors
    In NSDI, 2024
  3. Daman Arora*, Anush Kini*, Sayak Ray Chowdhury, and 3 more authors
    2023
  4. Anush Kini, Harish PB, Monica Anand, and 1 more author
    The Open Public Health Journal, 2023