Member of Technical Staff — RL & Agentic Model Training
Core
Build reinforcement learning infrastructure and post-training pipelines for multimodal, vision-language, and code-generating agents to improve their planning, tool use, and complex workflow execution.
Role type
Senior IC machine-learning engineer (RL & agentic model training)
Builds
RL and post-training pipelines, distributed training systems, and high-throughput rollout generation infrastructure
Domain
Robotics, Embodied AI, Interactive 3D Worlds, World Models
Deliverable
production ML models
Required skills
Reinforcement learning, post-training, large-scale fine-tuning, distributed training, high-throughput inference, supervised fine-tuning, preference optimization, reward modeling, agentic RL, code-generation agents, long-horizon evaluation, tool-using systems, Python, PyTorch, JAX
Preferred skills
Frontier AI lab experience, code-model post-training, autonomous coding agents, multimodal models, robotics, simulation, embodied AI, verifiable rewards, outcome-based training systems, scaling RL workloads
Technologies
PyTorch, JAX, multi-GPU environments
Responsibilities
Build RL and post-training pipelines for multimodal agents; Develop infrastructure for SFT, preference optimization, reward modeling, and RL; Create systems for collecting, filtering, replaying, and learning from agent trajectories; Design rewards, verifiers, and evaluations for long-horizon tasks; Scale distributed training and high-throughput rollout generation; Improve training reliability, reproducibility, and cost efficiency; Define long-term strategy for agent and embodied-model training
Seniority
Senior, hands-on IC
