Harikrishna Narasimhan
Staff Research Scientist at Google Deepmind
Based in Mountain View, United States
7-day free trial · no credit card
Seniority
Staff
Department
Science
Location
Mountain View
Industry
Research Services
Company size
9.9K
Contact information
Reveal Harikrishna's email and phone
Direct contact data is gated. Sign up and reveal. You only pay for verified records.
h•••••••@deepmind.google
Phone
5 credits+1 ••• •••• ••••
You only pay for valid records. Bounced emails and disconnected numbers cost nothing.
Background
About Harikrishna Narasimhan
My research is focused on improving inference efficiency in large language models (LLMs). Our recent paper on this topic (Narasimhan et al,"Faster cascades via speculative decoding", 2025) got an Outstanding Paper Honorable Mention Award at ICLR 2025.I also work more broadly on both applied and theoretical aspects of supervised machine learning (ML), including the design of loss functions for ranking and retrieval, constrained optimization for training ML models, and ML fairness.Our open-source library for constrained optimization can be found at: https://github.com/google-research/tensorflow_constrained_optimization/
Decision-makers
Other people at Google Deepmind
Build a list of verified contacts at Google Deepmind
Free for 7 days · 50 credits · no card · only pay for verified records.
Reach more buyers like Harikrishna
250M+ professionals with verified email and phone. You only pay for records that actually verify.
7-day trial · no credit card · cancel anytime