Teams at CHAI AI

Own AI model-serving infrastructure for a platform with 10 million users. Optimize model fit within fixed GPU budgets, improve serving speed and latency, manage cost per token, determine when to optimize or rebuild systems, and plan capacity expansion. The role requires high ownership, comfort with ambiguity, and in-person work at the Palo Alto office.

Own the post-training pipeline for AI roleplay models, transforming raw data into improved model behavior through supervised fine-tuning, reinforcement learning, reranking, data curation, and evaluation. The role involves closing the loop from evaluation signals to training runs, defining problems independently, and working onsite with a small, fast-moving engineering team.

Own the research agenda for AI model training and evaluation, including developing new training objectives and evaluation metrics. Connect research outcomes directly to revenue and product performance in a fast-moving social AI startup. The role requires high ownership, comfort with ambiguity, strong motivation, software engineering experience, and in-person work at the Palo Alto office.