Research Engineer — Post-Training, Mountain View, CA
Core
Develop and optimize model post-training methods (SFT, RLHF, GRPO, PPO, DPO) and build scalable training pipelines for large generative models to improve enterprise AI capabilities.
Role type
Senior Research Engineer (Post-Training)
Builds
Scalable training, rollout, and evaluation pipelines for large generative models
Domain
Enterprise AI, Machine Learning, Deep Learning
Deliverable
production ML models
Required skills
Model post-training (SFT, RLHF, GRPO, PPO, DPO), Deep learning, Reinforcement learning, Optimization, Python, PyTorch/JAX, GPU-based training pipeline optimization, Experimental design and analysis
Preferred skills
Original ML research contributions, Publications at NeurIPS/ICML/ICLR, Distributed training, Large-scale reinforcement learning, Reward modeling, Synthetic training data, Agentic RL
Technologies
PyTorch, JAX, DeepSpeed, Ray, verl, TRL
Responsibilities
Develop and optimize post-training methods; Design training objectives and reward functions; Build scalable training pipelines; Design rigorous experiments and analyze training dynamics; Optimize GPU-based training systems; Collaborate with research scientists to implement novel ideas
Seniority
Senior, hands-on IC
