Director, Reinforcement Learning & Agentic Post-Training

Tasks

• Build agent training and reinforcement learning environments
• Create training data generation and preference pipelines
• Design evaluation frameworks for multi step workflows
• Develop reward models and verifiers
• Improve policies using supervised fine tuning preference optimization and reinforcement learning
• Lead reinforcement learning strategy for LLM agents
• Measure tool call correctness and workflow completion
• Mentor machine learning engineers and review technical designs
• Partner with product and domain experts to build trainable agent environments
• Set engineering standards for experiment tracking and reproducibility

Perks/Benefits

• N/A

Skills/Tech-stack

AI Feedback | API Integration | Distributed Training | Environment Design | Evaluation | Experiment tracking | Fine Tuning | GRPO | Human Feedback | Language Models | Large Language Models | Learning from Human Feedback | Megatron | Megatron-LM | NEMO | NVIDIA Nemo | NVIDIA Nemotron | Observability | Offline Reinforcement Learning | PPO | Policy Optimization | Preference optimization | PyTorch | Python | Ray | Reinforcement Learning | Reinforcement Learning from AI Feedback | Reinforcement Learning from Human Feedback | Rejection Sampling | Reproducibility | Reward Modeling | Reward shaping | Rollback Safety | Supervised Fine Tuning | Tool use | VLLM