Nikhil Barhate
Machine Learning Research Engineer at Scale AI
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolNikhil Barhate is a Machine Learning Research Engineer based in San Francisco with eight years of experience building and scaling RL and LLM systems from research to production. He has designed distributed training and inference frameworks for large-scale reinforcement learning and reasoning language models at Scale AI, and previously advanced hierarchical and multimodal transformer approaches during graduate research at CU Boulder. His work spans applied research and engineering—deploying judge LLMs for data quality, optimizing CPU-GPU partitioning for AMD accelerators with a 24% benchmark gain, and contributing a compact PyTorch PPO implementation that reflects hands-on RL expertise. Academically strong (MS CS, 3.97 GPA) with a solid electronics engineering foundation, he blends deep algorithmic insight with pragmatic system design to accelerate ML experimentation and deployment.
8 years of coding experience
1 year of employment as a software developer
Master of Science, Computer Science, 3.97 / 4.00, Master of Science, Computer Science, 3.97 / 4.00 at University of Colorado BoulderBachelor of Technology, Electronics Engineering, 9.09 / 10.00, Bachelor of Technology, Electronics Engineering, 9.09 / 10.00 at University of Mumbai
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (6)
pytorch10
ppp10
reinforcement-learning9
deeplearning-ai9
deep-learning9
deep-reinforcement-learning9
Programming languages (3)
Github contributions (5)
nikhilbarhate99/PPO-PyTorch
Sep 2018 - Jan 2023
Minimal implementation of clipped objective Proximal Policy Optimization (PPO) in PyTorch
Role in this project:
ML Engineer
ML EngineerContributions:89 commits, 8 PRs, 87 pushes in 4 years 4 months
Contributions summary:Nikhil primarily contributed to the implementation of Proximal Policy Optimization (PPO) algorithm within a PyTorch framework. Their commits showcase the development and modification of a `Model` class, including the definition of actor and critic networks, the handling of action distributions, and the overall structure of the PPO algorithm. The changes involve iterating on the PPO implementation, including updates to memory management and the policy update steps, demonstrating a focus on reinforcement learning techniques.
pytorchpolicyreinforcement-learning-algorithmspytorch-implmentionpolicy-optimization
nikhilbarhate99/Actor-Critic-PyTorch
Sep 2018 - May 2019
Contributions:57 commits, 46 pushes, 1 branch in 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.