Kislay Aditya Oj
Computer Science and Engineering | MS · IIT Bombay | BTech · IIT Kanpur
About
I'm Kislay, currently pursuing an MS by Research in Computer Science & Engineering at IIT Bombay, where I'm part of the CFILT Lab. I completed my B.Tech from IIT Kanpur in 2025. These days, my primary focus is on inference time algorithms and optimizations. I also have a strong interest in reinforcement learning, machine learning and language technologies, and enjoy working on problems that involve learning theory, decision making and statistics.
Outside of academics, I really enjoy chess, I follow top games closely and play competitively (currently around 1750, though I have touched 2000 before… we don't talk about the rating drop, also it's chess com rating not FIDE, i'm not a genius). I also follow Formula 1, sketch in my free time, and enjoy reading fiction. Lately, I've been learning to play the piano as well. You can find me on X, check out my occasional blogs in the notes section, browse through my projects, or see my vibe projects.
Research
Efficient AI
I work on making large models cheaper and faster to run at inference time. My current threads are KV cache compression for language models and sparse video generation. The broad goal across both is to cut computation and memory costs while keeping the methods verifiable, so the speedups come with clear evidence that output quality is preserved rather than just a claim.
Reinforcement Learning for Adaptive LLM Agents
I study how reinforcement learning ideas can be used to help large language model (LLM) agents improve from feedback during real use. This work focuses on test time adaptation using reflection, memory, and personalization rather than model fine tuning. As part of the Flipkart–IIT Bombay collaboration, my goal is to build LLM agents that become more reliable and user aware over time.
Other Interests
Beyond my main threads, I'm interested in mechanistic interpretability and am working on a few projects around understanding what goes on inside models. I'm also part of the Bhashini initiative, where I work on training machine translation systems. More recently, I've been exploring agents that generate benchmarks automatically using adversarial techniques.
Publications
working on it
Contact
Email: oasiskislay@gmail.com · kislay@cse.iitb.ac.in
Lab: CFILT Lab, Room 401, Computer Centre
Department of Computer Science
Indian Institute of Technology, Bombay
Google Scholar · GitHub · Twitter · LinkedIn
Loading notes...
Loading projects...
Loading vibe projects...