Senior AI Engineer, Post-Training

What you'll do

• Post-train open-weight language models on proprietary legal data using supervised fine-tuning and reinforcement learning.
• Manage the end-to-end model development lifecycle including data selection, objective design, training, and evaluation.
• Build and improve high-quality training datasets and automated data pipelines with human feedback loops.
• Maintain and optimize the training stack to ensure reliable experiments on managed or self-hosted infrastructure.
• Develop and operate production systems for model serving, agentic workflows, and evaluation pipelines.
• Collaborate with product engineers to decide between model improvements and system-level enhancements like tools and context.
• Translate complex legal workflows into technical requirements for model training and data strategy.

What we're looking for

• Experience with LLM post-training using PyTorch or equivalent frameworks.
• Experience in model development and post-training work in applied settings that shipped to real users.
• Ability to build and operate production systems including model serving, agents, and evaluation pipelines.
• Proficiency in training techniques such as supervised fine-tuning, preference optimization, and reinforcement learning.
• Ability to develop training datasets and data pipelines involving labeling guidance and human feedback loops.
• Knowledge of distributed training to diagnose and optimize training runs on managed or self-hosted infrastructure.
• Ability to translate complex domain expert workflows into model, data, and evaluation decisions.