Member of Technical Staff
Core
Design and deploy post-training pipelines for large language models, focusing on supervised fine-tuning, reinforcement learning, and reward model development to enable tool use and agentic workflows.
Role type
Senior Machine Learning Engineer (Post-Training & RL)
Builds
Production-ready LLMs with agentic capabilities and tool-use functionality
Domain
Artificial Intelligence / Large Language Models / Reinforcement Learning
Deliverable
production ML models
Required skills
Python, PyTorch, JAX, Reinforcement Learning (SFT, RLHF, DPO, GRPO), Distributed Training (FSDP, DeepSpeed, Ray, Kubernetes, Slurm), Reward Model Design, Model Evaluation
Preferred skills
Agentic/Tool-use training, Synthetic data generation, Retrieval-augmented generation, Academic publications (NeurIPS, ICML, ICLR, MLSys, ACL, EMNLP)
Technologies
PyTorch, JAX, FSDP, DeepSpeed, Ray, Kubernetes, Slurm
Responsibilities
Create high-quality training datasets from prompts and responses; Build and iterate on supervised fine-tuning and reinforcement learning pipelines; Design reward signals using reward models and LLM judges; Train models for tool use and long-horizon agentic workflows; Define model evaluations and assess training runs; Prototype post-training methods and move them to production; Deploy trained models and own their production performance.
Seniority
Senior, hands-on IC