2027 Internship Behavior ML Engineer, Learned Manipulation Policies
Bedrock Robotics Inc9 hours ago
Responsibilities
- Train and iterate on diffusion flow-matching and related behavior policies using fleet demonstration data and simulated rollouts.
- Run architecture, conditioning, and hyperparameter experiments and communicate clear findings.
- Build and improve evaluation pipelines for task success, smoothness, safety margins, and operator-likeness.
- Investigate distribution shift, mode collapse, and out-of-distribution scenes and propose fixes.
- Collaborate with simulation and evaluation teams to connect offline metrics with on-machine performance.
- Contribute clean, reviewed, and reproducible work to the shared training codebase.
- Share results with behavior, controls, and autonomy teams.
Requirements
- Currently pursuing a BS, MS, or PhD in computer science, robotics, machine learning, or a related field, or bringing equivalent research or industry experience.
- Strong Python skills and hands-on experience training models in PyTorch or an equivalent framework.
- Working knowledge of generative modeling, diffusion models, flow matching, VAEs, imitation learning, or behavior cloning.
- Experience running and interpreting real training experiments.
- Ability to clearly explain experiments, results, and next steps.
- Preferred: published work or substantial project experience in diffusion, flow matching, robot learning, or imitation learning.
- Preferred: experience with Vision Language Action models, reinforcement-learning policies for robot manipulation, policy evaluation in simulation, sim-to-real reasoning, large-scale training infrastructure, or experiment tracking tools.
- Preferred: exposure to robotics, autonomous vehicles, or other physical-world control problems.
Categories
ML EngineeringRobotics
About Bedrock Robotics Inc
Bedrock Robotics builds autonomous control systems that retrofit heavy construction equipment, enabling driverless operation on large infrastructure and industrial projects. The San Francisco–based, privately held company, founded in 2024, deploys its technology with contractors and project owners to speed schedules and improve job-site safety. Its business centers on upgrading existing fleets and operating them in the field, with software, sensors, and integration services tailored to construction workflows.