Bo Wang is a Co-Founder and CTO in the San Francisco Bay Area who builds and scales research-driven product teams that turn 3D vision research into practical AIoT systems. He led Deptrum’s R&D of 50+ engineers and shipped innovations such as a sub-$100 ToF-based 3D SLAM product, DepthMagic (a stereo-depth algorithm reportedly 100x faster than OpenCV), and the sToF camera that fuses structured-light with time-of-flight for longer range and lower power. His background spans deep learning systems at Zoox and hands-on ML infra work—he contributed an ELU converter and INT64→INT32 auto conversion to the high-profile pytorch/TensorRT project—showing fluency from algorithms to runtime optimization. With training from Tsinghua and doctoral studies at Stanford, he combines hardware-aware sensing inventiveness with software and compiler-level optimizations to deliver energy-efficient, product-ready solutions. He’s known for turning hard research problems into scalable products and for bridging the gap between academic rigor and shipping engineering.
PyTorch/TorchScript/FX compiler for NVIDIA GPUs using TensorRT
Role in this project:
ML Engineer
Contributions:82 reviews, 146 commits, 83 PRs in 1 year 9 months
Contributions summary:Bo contributed to the PyTorch/TorchScript/FX compiler for NVIDIA GPUs using TensorRT repository by creating and compiling an ELU converter. They implemented the ELU activation function within the TensorRT environment and created a test to validate it, showing a focus on integrating and testing custom operations. Further contributions included refactoring the example directory and refactoring the ELU converter, including implementing the auto type conversion for INT64 to INT32.
Contributions:88 pushes, 1 branch in 3 years 2 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.