Eric Undersander
Software Engineer at Stealth Venture
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolEric Undersander is a software engineer with two decades of experience building high-performance robotics, simulation, and ML systems, most recently as a Research Engineer on Meta FAIR’s robotics teams and now at a stealth venture in San Francisco. He blends low-level C++ performance optimization, multithreaded deterministic algorithms, and computational geometry from autonomous vehicle and game-engine work with ML systems engineering for training and profiling large models. At Cruise he led technical direction for motion and behavior planning, translating behaviors into convex optimization and production-grade graph search, while at Baidu he accelerated DNN training and built GPU profilers that delivered ~10x speedups. His open-source contributions to Facebook Research’s Habitat-Sim and Habitat-Lab emphasize profiling, performance tooling, and simulation infrastructure critical to embodied AI research. Comfortable across research and product settings, he pairs deep systems-level thinking with practical instrumentation to squeeze out predictable, measurable performance. An uncommon throughline in his career is turning profiling and tracing insights into tooling that improves both developer workflow and runtime efficiency.
9 years of coding experience
23 years of employment as a software developer
BS Computer Science, BS Computer Science at The University of Texas at Austin
Stackoverflow
366reputation
22kreached
6answers
0questions
Github Skills (26)
pytorch10
performance-monitor10
c-language10
performance-analytics10
code-profiling10
python10
reinforcement-learning10
performance-measurement10
performance-analysis10
ai10
profiling10
performance-tuning10
deep-reinforcement-learning10
performance-monitoring10
cprogramming-language10
Programming languages (5)
Github contributions (5)
facebookresearch/habitat-sim
Aug 2020 - Oct 2022
A flexible, high-performance 3D simulator for Embodied AI research.
Role in this project:
Backend & Performance Engineer
Backend & Performance EngineerContributions:483 reviews, 180 commits, 184 PRs in 2 years 2 months
Contributions summary:Eric contributed significantly to the development of profiling and performance optimization tools within the Habitat-Sim project. This included the creation of `profiling_utils.py` to annotate code sections and `compare_profiles.py` to analyze profiling data. The user also refactored and improved existing code, such as updating the color buffer and fixing issues. The impact of these contributions enhances the project's performance analysis capabilities.
roboticsembodied-aicomputer-visionsimulation3d-simulator
facebookresearch/habitat-lab
Aug 2020 - Oct 2022
A modular high-level library to train embodied AI agents across a variety of tasks and environments.
Role in this project:ML Engineer & Performance Engineer
ML Engineer & Performance EngineerContributions:357 reviews, 86 commits, 85 PRs in 2 years 2 months
Contributions summary:Eric primarily contributed to the integration and utilization of profiling tools within the repository. They introduced profiling configurations and ranges for PPO and DDPPO training, alongside implementing a profiling wrapper, thereby enabling performance analysis. Furthermore, they removed an older version of profiling utilities from the main repository and added a script for generating profiling shell scripts, demonstrating a focus on optimizing and measuring the performance of the training pipelines. This likely improved the efficiency of model training within the Habitat-lab environment.
pythonembodied-aideep-reinforcement-learningsimulatorai-agents
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.