Kunal Bhalla
Member Of Technical Staff at Thinking Machines Lab
New York, New York, United States
Summary
🤩
Rockstar🎓
Top SchoolKunal Bhalla is a seasoned software engineer with 15 years of experience building developer-facing infrastructure and ML tooling, most recently as a Principal Software Engineer at Meta and now a Member of Technical Staff at Thinking Machines Lab in New York. He has a strong track record in debugging tools and AI infrastructure—shifting large organizations from product engineering to ML developer productivity and contributing to open-source ML ecosystems like PyTorch, where he improved torch.fx wrapping, pickle support, and distributed-run reliability. Early work spans mobile reliability, battery instrumentation (open-sourced at Facebook), and systems that reduced pipeline latencies by orders of magnitude, reflecting a pragmatic focus on observability and performance. Trained as an engineer at IIT Delhi, he combines deep systems instincts with hands-on contributions to widely used ML frameworks, and maintains a personal programming site at knlb.dev.
15 years of coding experience
14 years of employment as a software developer
Indian Institute of Technology Delhi (IIT Delhi)High School, High School at Bal Bharati Public School PitampuraDelhi Public School - R. K. Puram
Github Skills (13)
neural-network10
machine-learning10
pytorch10
deep-learning10
python10
autograd10
testing9
gpu9
package-management9
tensor9
debug8
debugging8
distributed-computing8
Programming languages (13)
Github contributions (5)
pytorch/pytorch
Oct 2020 - Sep 2022
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Role in this project:
ML Engineer
ML EngineerContributions:12 reviews, 5 commits, 11 PRs in 1 year 11 months
Contributions summary:Kunal's contributions focused on enhancing the `torch.fx.wrap` functionality and improving the package handling capabilities within the PyTorch framework. They addressed a bug related to decorator handling and the incorrect storage of function names. The user also implemented changes to support pickle protocol 4+ and switched tests to use buffers, streamlining the save/load test suite. Furthermore, the user made adjustments to the distributed run utility, and corrected an issue where device indices were improperly logged.
pythongpu-accelerationdeep-learninggpunumpy
kunalb/EventPress
May 2011 - Dec 2011
Contributions:34 commits in 7 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.