Post-Training AI Researcher: Production RL & LLM Tuning

Baseten; San Francisco (CA); On-site; USD 200,000 - 275,000; Full time

Benefits offered by this job

Equity

Medical/Dental/Vision insurance

Flexible PTO with Winter Break

Parental leave

Fertility stipend

401(k)

Learning opportunities

Job summary

Baseten is at the applied end of post-training research, partnering with the world’s fastest-growing AI companies to post-train open-source models that outperform frontier closed models on their specialized tasks. Your day-to-day is finding signal in complex datasets and building the training pipelines that turn that signal into production-ready models.

We’re seeking hands-on researchers who ship models, translate customer requirements into curricula, and iterate quickly while staying rigorous.

Qualifications

• Hands-on experience training LLMs with reinforcement learning.
• Translate customer needs into effective training curricula.
• Build multi-turn environments with tool use and agentic workflows.

Responsibilities

• Design and run post-training pipelines: SFT, GRPO, DPO, RLVR, reward function engineering, and synthetic data generation.
• Build task-specific training environments and evals for healthcare, code generation, and legal domains.
• Collaborate with customers to translate production data into training signals and reward loops.
• Run and analyze training experiments end-to-end; diagnose reward hacking and drift.
• Publish findings and contribute to Baseten's open-source training libraries.

Skills

LLM fine-tuning with RL

Reward engineering

Multi-turn environments

Production ML systems

Experimentation & debugging training

Tools

RL training frameworks

Job description

Baseten is at the applied end of post-training research, partnering with the world’s fastest-growing AI companies to post-train open-source models that outperform frontier closed models on their specialized tasks. Your day-to-day is finding signal in complex datasets and building the training pipelines that turn that signal into production-ready models.

We’re seeking hands-on researchers who ship models, translate customer requirements into curricula, and iterate quickly while staying rigorous.