Rishi Puri
Engineering Manager at NVIDIA
Berkeley, California, United States
Summary
🤩
RockstarRishi Puri is an Engineering Manager based in Berkeley, CA with nine years of experience leading and shipping performant machine learning infrastructure. As an Engineering Lead and core contributor to PyTorch Geometric at NVIDIA, he has hands-on expertise optimizing graph neural network kernels—integrating pyg_lib.segment_matmul into RGCNConv and HeteroLinear and improving heterogeneous graph processing and numerical stability across tests. He blends backend engineering and ML research, routinely turning algorithmic improvements into production-grade library enhancements that accelerate training and inference. Known for pragmatic problem-solving, he focuses on hotspots that unlock order-of-magnitude speedups rather than incremental tweaks. Colleagues rely on him to bridge deep technical contributions with team-level delivery and long-term maintainability.
9 years of coding experience
Spanish, English
Github Skills (12)
pytorch10
machine-learning10
graph-neural-network10
python10
linear-algebra10
gnn10
optimization9
deep-learning9
algorithms8
data-structures8
algorithm8
data-structure8
Programming languages (4)
Github contributions (5)
pyg-team/pytorch_geometric
Jul 2022 - Jan 2023
Graph Neural Network Library for PyTorch
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:201 reviews, 425 commits, 199 PRs in 5 months
Contributions summary:Rishi made significant contributions to the `pytorch_geometric` library by integrating `pyg_lib.segment_matmul` functionality into the `RGCNConv` and `HeteroLinear` layers, accelerating graph neural network computations. They also worked on the `HGTConv` and added features to the `to_homogeneous` method, demonstrating a focus on optimizing and extending the library's core functionalities for heterogeneous graph processing. Furthermore, the user addressed issues related to the usage of `pyg_lib` and improved the numerical stability of several testing procedures.
pytorchgraph-convolutional-networksgeometric-deep-learningdeep-learningneural-graph
puririshi98/pytorch
May 2021 - Feb 2023
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Contributions:207 pushes, 10 branches in 1 year 9 months
pythongpu-accelerationdeep-learninggpuacceleration
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.