My current research is on continual learning: how a foundation model can learn from new data without being retrained on an updated data mixture. I am interested in the theory behind this, and in applying it to scientific discovery, where new results keep coming in.
Keywords: Continual Learning, Foundation Models, Reinforcement Learning and Agents
For more details about my background, refer to my Resume / CV. I am always happy to discuss research, potential collaborations, or just chat about ML. Feel free to reach out with a specific topic of discussion via Email or LinkedIn!
Last Updated: October 2026
-
October, 2026
Started my PhD at the University of Oxford, in the OATML group.
-
August, 2026
Gave a talk and presented a poster on goal coverage in RL at RLC (Montreal).
-
August, 2026
Joined Proximal to build long-horizon RL environments.
-
December, 2025
Presented work at Google DeepMind's NeurIPS Booth on goal coverage in RL (San Diego).
-
October, 2025
Our paper "Dense and Diverse Goal Coverage in Reinforcement Learning" is now on arXiv! [Paper]
-
July, 2025
Our work on "Learning from Label Proportions and Covariate-shifted Instances" has been accepted to UAI (Rio)! [Paper]
-
December, 2024
Our paper "Learning from Label Proportions and Covariate-shifted Instances" is now on arXiv! [Paper]
-
August, 2024
Joined Google DeepMind as a Pre-Doctoral Researcher.
-
July, 2024
Oracle's flagship TTS model supporting natural voices and voice cloning is live.
-
August, 2023
Joined Oracle Cloud AI Services - Speech team.
-
May, 2023
Gave oral presentation at AAMAS about our work on combining human-AI predictions (London).
-
May, 2023
Graduated from IIT Ropar with a CGPA of 9.31/10 (AI Concentration: 10/10)
-
May, 2023
Awarded the AAMAS Student Scholarship (£650), Google Research Travel Grant ($2k) and Microsoft Research Travel Grant (₹120k).
-
February, 2023
Offered Amazon Applied Scientist Internship based on performance in Amazon ML Summer School and interviews (Alexa team).
-
December, 2022
Top 20 in India in Oppo Inspiration Cup; invited to finale in Hyderabad.
-
August, 2022
Finished SWE internship at Oracle with a full-time return offer.
-
May, 2022
Received Best B.Tech. Project Award during Technology Day at IIT Ropar.
-
May, 2022
Achieved global rank 238 in Google Kickstart Round C.
-
March, 2021
Achieved All India Rank 18 and Global Rank 198 in Google Hash Code.
Verifiable Problem Generation for Gemini Thinking
Engineered a fully-verifiable curriculum generation engine for improving
thinking in Gemini models.
Contributed 38% of the data in the Gemini thinking data revision, leading to >10%
improvement on downstream tasks like Math and NL Reasoning.
Transformer-based Text-to-Speech (TTS)
Built Oracle's in-house TTS model supporting few-shot voice cloning.
Handled high-fidelity data procurement, distributed training on Slurm clusters,
deployment via Triton inference server, and developer experience.
Blog:
OCI Speech Update
SAMPAN: Tracking District Level Malnutrition
Developed an Android app and Web dashboard used by 1200+ workers to log 50,000+ data points
monthly.
Product:
Play Store
Video
Media Coverage:
The Tribune
C=Conference, P=Preprint, *=Equal Contribution
