Stanford, California, United States
ML engineer with 7+ years spanning the full robotics and autonomy stack—from data collection and infrastructure to model training, on-device deployment, and reinforcement learning in the real world. Equally at home setting research direction and shipping production systems, with a track record of taking deep learning from prototype to deployment at scale.
Teaching robots to teach themselves
- One of the youngest technical leads; owned camera blindness detection and operational design domain sensing end to end—from stakeholder requirements through delivery of a state-of-the-art, safety-certified system shipping in production vehicles in 2025. - Built the full production deep learning stack—data curation, training, quantization, evaluation, and efficient in-car deployment—and authored a novel training method that detected blockages faster and more completely than prior art. - Primary author of the initial 3D lane detection training pipeline (working prototype in just 3 days); widely reused as the foundation for later 3D detection networks. - Wrote safety-compliant C++/CUDA for DNN deployment and post-processing. - Ranked a Top Contributor every year and among the top 5 of 200+ engineers in the perception organization.
Built automation infrastructure streamlining DNN development, KPI computation, and deployment with full experiment traceability, plus a data-mining tool to surface mislabeled and challenging examples from unlabeled datasets.
- Developed several methods for monitoring the outputs of neural networks for direct use in pedestrian detection monitoring. The resulting paper was accepted to IEEE Intelligent Transportation Systems Conference 2020. - Collaborated with other researchers to develop a system for building a Dynamic Scene Graph (DSG) of a building automatically, the first system of its kind. The resulting paper was accepted into Robotics Science an Systems (RSS) 2020, and it will be presented as a workshop paper at Computer Vision and Pattern Recognition (CVPR) 2020. - Developed a first attempt at verifying a full mesh of the environment by predicting per-face error using a graph neural network which operates on the dual of the environment mesh.
- Developed a sensitivity analysis pipeline to assess neural network weaknesses - Helped improve path perception neural networks through the addition of augmentations and some minor architecture changes.
• Collaborated with two postdoctorates to develop a pedestrian trajectory prediction model that integrates information from the environment • Integrated a model architecture from a top paper in the field • Refactored codebase for clarity and efficiency