2027 Internship Behavior ML Engineer, Learned Manipulation Policies

What you'll do

• Train and iterate on diffusion flow-matching and related behavior policies; explore architectures, conditioning, and hyperparameters.
• Build evaluation pipelines for task success, smoothness, safety margins, and operator-likeness; investigate policy failure modes.
• Collaborate with simulation, evaluation, behavior, controls, and autonomy teams, and contribute reproducible work to the training codebase.

What they ask for

• Currently pursuing a BS, MS, or PhD in computer science, robotics, machine learning, or a related field, or equivalent research or industry experience.
• Strong Python skills and hands-on model training experience in PyTorch (or equivalent).
• Working understanding of generative modeling (such as diffusion models, flow matching, or VAEs) and imitation learning or behavior cloning.
• Experience running and interpreting model-training experiments; clear communication.

Nice to have

• Published work or substantial project experience in diffusion, flow matching, robot learning, or imitation learning; experience with Vision Language Action (VLA) models or Reinforcement Learning (RL) policies for robot manipulation.
• Experience evaluating policies in simulation, reasoning about the sim-to-real gap, or using large-scale training infrastructure and experiment tracking tooling; exposure to physical-world control.

Tools

Python, PyTorch, diffusion models, flow matching, VAEs, Vision Language Action (VLA), Reinforcement Learning (RL)

AI focus

Trains and evaluates generative behavior policies for robot manipulation, investigating failure modes and the relationship between simulation metrics and on-machine performance.