Prakash Prabhu
Staff ML Compiler Engineer at Waymo
Bengaluru, Karnataka, India
Summary
👤
Senior🎓
Top SchoolPrakash Prabhu is a Staff ML Compiler Engineer with 10+ years of experience building high-performance ML compilation and inference systems for TPUs across edge and data-center deployments. He has led and grown compiler teams, designed front-to-back MLIR-based compiler features, and pioneered 4-bit quantized model pipelines and pipeline-parallel execution for real-world products like Gemini Nano and ASR. His work blends compiler research (multi-core auto-parallelization, auto-tuning with RL and simulation) with production engineering, earning multiple Google-wide performance and tech impact awards and publications at MLSys and ICML. At Waymo he is the founding engineer for the ML Compiler team in Bangalore, bringing deep expertise in partitioning and multi-TPU orchestration. An active open-source contributor, he has improved core probabilistic modeling code in tensorflow/probability to harden edge cases and tests. He pairs a Princeton PhD and research background in exposing implicit parallelism with a pragmatic record of shipping latency-sensitive ML features on mobile and cloud hardware.
10 years of coding experience
19 years of employment as a software developer
Masters in Engg, Computer Science, Masters in Engg, Computer Science at Indian Institute of ScienceB.E., Computer Science, B.E., Computer Science at National Institute of Technology TiruchirappalliDoctor of Philosophy (PhD), Computer Science, Doctor of Philosophy (PhD), Computer Science at Princeton University
Github Skills (6)
machine-learning10
debug10
probabilistic-programming10
tensorflow10
python10
statistics9
Programming languages (1)
Github contributions (5)
tensorflow/probability
Jul 2019 - Aug 2019
Probabilistic reasoning and statistical analysis in TensorFlow
Role in this project:
ML Engineer
ML EngineerContributions:29 commits, 1 PR, 12 comments in 1 month
Contributions summary:Prakash primarily contributed to the `tensorflow/probability` repository by fixing bugs and making improvements to the `LinearGaussianStateSpaceModel` class, which is a core component for probabilistic modeling. These changes included addressing issues with the model's behavior when the number of time steps was one, fixing type errors, and handling potential issues with empty tensors. Additionally, the user merged code from a different repository, suggesting a collaborative workflow. The contributions indicate an effort to improve the stability and reliability of the model, including test updates.
statisticspythonprobabilistic-reasoningdata-sciencedeep-learning
prakashkagitha/pro
Nov 2015 - Jun 2018
Contributions:1 push, 1 branch in 2 years 8 months
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.