2027 Summer Intern, PhD, Research, Behavior Post-training

What you'll do

• Build tools to sample, select, and discard synthetic driving data using testable improvement hypotheses.
• Train and evaluate models using mixtures of synthetic and logged data.
• Present and write about synthetic data generation and training methods.

What they ask for

• Currently pursuing a PhD in Computer Science, Machine Learning, Robotics, Electrical Engineering, or a related technical field.
• Familiarity with methods for sampling from large models.
• Experience with post-training methods such as SFT, RL, and preferences.
• Experience with modern deep learning frameworks such as PyTorch and Jax.

Nice to have

• Experience with open- and closed-loop decision-making models in robotics, and with synthetic data generation and evaluation.
• Prior projects using LLM-as-a-judge.

Pay

$85 — $85 USD (hourly pay)

Tools

PyTorch, Jax, SFT, RL, LLM-as-a-judge

AI focus

Machine learning post-training for autonomous driving, including synthetic data generation and evaluation, sampling from large models, and training with synthetic and logged data.