PhD Research Intern – Learned End-to-End Driving and Actuation

About the role

The Perception and Machine Learning Team is seeking a final‑year PhD student to join as a Research Intern focused on learned planning and actuation for autonomous vehicles. The internship is on‑site in Berkeley, CA and may lead to a full‑time research position.

Key responsibilities

• Own a research direction on the production motion planner, working with researchers and real fleet data.
• Identify failure mechanisms in the planner, propose and evaluate improvements, and present results to decision‑makers.
• Develop and test generative planning, long‑tail data balancing, closed‑loop evaluation, and post‑training reinforcement learning methods.
• Evaluate changes both open‑loop and closed‑loop, analyze failures, and support publication of non‑proprietary findings.

Required profile

• Final‑year PhD student (or within 12 months of defending) in computer science, electrical engineering, robotics or related field.
• First‑author research in motion forecasting, planning, generative models, reinforcement or imitation learning, or closed‑loop simulation.
• Strong communication and teamwork skills; ability to brainstorm and share ideas openly.

Required skills

• Python programming.
• PyTorch deep‑learning framework.

What we offer

• Direct impact on a product with a defined path to a full‑time research role.
• Dedicated multi‑GPU cluster, access to fleet data, labels and evaluation pipeline from day one.
• 40‑hour weeks for 3–6 months, flexible start date, compensation of $55.00 /hr.
• Daily catered lunches, fully stocked fridge, and a venture‑backed multidisciplinary team.