About the role#
This is a full-time, paid internship for Summer 2027. You will work in a hybrid capacity based in Mountain View, California. You will join our team to build and test agentic AI systems that advance autonomous driving research.
What you'll need#
- Current enrollment in a PhD program for Computer Science or a closely related field.
- A track record of independent research in machine learning, reinforcement learning, agentic AI, or human-machine interaction.
- Deep technical experience developing autonomous agentic workflows, recursive self-improvement loops, or LLM orchestration frameworks.
- Experience with publications in top-tier venues like NeurIPS, ICML, ICLR, CHI, or CVPR is preferred.
- Practical experience with self-play architectures, auto-research loops, or recursive self-improvement loops is preferred.
- Experience with advanced reinforcement learning techniques, such as offline RL, RLHF/RLAIF, or curriculum learning in complex simulation environments, is preferred.
What you'll do#
- Architect and implement end-to-end autonomous research workflows using Google Antigravity. These workflows allow LLM-powered agents to generate, run, and evaluate post-training experiments.
- Build and refine agentic feedback mechanisms. These help reinforcement learning models learn from simulation rollouts and self-correct.
- Set up evaluation and root-cause analysis pipelines to monitor, debug, and improve the decision-making quality of autonomous research agents.
- Collaborate with senior researchers to document methodology, benchmark performance gains, and prepare academic conference submissions.
Location & details#
- Location: Mountain View, CA
- Term: Summer 2027
- Modality: Hybrid
- Compensation: $85 per hour


