Post-Training Applied Researcher
Core
Design and run post-training pipelines (SFT, GRPO, DPO, RLVR) to fine-tune open-source LLMs for specialized customer tasks using domain-specific datasets and reward engineering.
Role type
Applied Researcher (LLM Post-Training & RL)
Builds
Production-ready LLMs and training pipelines for customers in healthcare, code generation, and legal domains.
Domain
Artificial Intelligence / Large Language Models / Reinforcement Learning
Deliverable
production ML models
Required skills
LLM fine-tuning, Reinforcement Learning (GRPO, PPO), Reward Engineering, Multi-turn agent environment design, Synthetic data generation, Training pipeline orchestration, Experiment analysis (reward hacking, drift)
Preferred skills
RL training frameworks, Publications at NeurIPS/ICML/ICLR
Technologies
SFT, GRPO, DPO, RLVR, Python, PyTorch, Hugging Face (implied by open-source context)
Responsibilities
Design and run post-training pipelines including SFT, GRPO, DPO, RLVR, and reward function engineering. Build task-specific training environments and evals for customer domains. Work with customers to translate production data into training signal and design reward loops. Run and analyze training experiments to diagnose instabilities. Publish findings at top venues.
Seniority
Mid-Senior, hands-on IC
