Tianshu Wang is a Senior Director based in Beijing with a PhD in Computer Vision & AI and over two decades of hands-on product development across client and business software, consumer hardware, robotics, smart cameras, IoT and IC development. He combines deep research pedigree in computer vision, graphics and machine learning with 10+ years focused on HCI and user experience, shaping products from mobile OS and groupware to intelligent camera systems. At Megvii he leads cross-disciplinary teams, translating advanced AI research into deployable products for real-world sensing and robotics applications. His practical engineering contributions extend into open source—fixing logging and checkpointing behaviors in PyTorch Lightning and refining Emacs keybindings—showing a willingness to improve infrastructure and developer ergonomics. Colleagues describe him as a bridge between rigorous academic methods and pragmatic product execution, comfortable moving between low-level IC/robotics concerns and high-level UX strategy. He brings a rare combination of long-term systems experience and hands-on ML tooling work that accelerates both research and production readiness.
10 years of coding experience
9 years of employment as a software developer
PhD, Computer Vision & AI, PhD, Computer Vision & AI at Xi'an Jiaotong University
Contributions:3 reviews, 10 commits, 10 PRs in 10 months
Contributions summary:Tianshu primarily contributed to the `evil-collection` repository, which provides keybindings for the Emacs editor. Their work focused on adapting and updating keybindings for different modes and packages like Vertico and Mu4e within Emacs. The commits involve modifying configuration files and adapting functionality, often related to handling user input through Evil-mode. The user also added and modified keybindings for other Emacs packages.
Pretrain, finetune ANY AI model of ANY size on multiple GPUs, TPUs with zero code changes.
Role in this project:
ML Engineer
Contributions:27 reviews, 8 commits, 19 PRs in 1 year 4 months
Contributions summary:Tianshu primarily contributed to the `pytorch-lightning` repository by addressing issues related to the WandbLogger. Their work involved fixing WandbLogger initialization, ensuring correct setting of `save_dir`, and correctly using the `run` name for logging. Additionally, the user made modifications related to saving checkpoints and profiler outputs, demonstrating involvement in experiment tracking and model management. These contributions are focused on enhancing the usability and functionality of the logging and checkpointing features within the PyTorch Lightning framework.
pythonheadachespytorch-modelsdata-sciencehandling
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.