Member of Technical Staff, Reinforcement Learning

6 months ago

San Mateo, CA, USAMid Level

Responsibilities

Design, develop, and optimize RL training pipelines for diffusion-based LLMs using PPO, DPO, RLHF, and novel approaches.
Build and iterate on reward models, reward-shaping strategies, and reward-quality evaluation methods.
Implement fine-tuning and scaling techniques for generative AI models.
Develop data preprocessing pipelines, model evaluation workflows, and alignment methods for enterprise use cases.
Research techniques for controlled text generation and constraint satisfaction.
Improve the stability, efficiency, and reproducibility of distributed RL workloads.

Bachelor's, master's, or PhD in Computer Science or a related field, or equivalent experience.
At least 2 years of experience working on ML projects using PyTorch or an equivalent framework.
Strong familiarity with transformers and core LLM concepts, including autoregressive pretraining, instruction tuning, and in-context learning.
Hands-on experience with RLHF, PPO, DPO, or related post-training methods.
Familiarity with training and inference in diffusion models.
Experience training deep learning models at scale in distributed computing environments.
Preferred qualifications include extensive experience training transformer-based language models from scratch, designing reward models or preference-learning systems, and knowledge of mixed precision, gradient accumulation, optimization theory, and neural network architecture design.
Experience with LLM serving frameworks such as vLLM, SGLang, or TensorRT is preferred.

51-200 employees