Agent Post-Training, API & Power Users

📋 Description Design and run experiments to improve API/power-user behavior Build evals, graders, environments from real workflows Turn observed failures into training data and improvements Improve model behavior when tools are used in API workflows Own end-to-end model behavior projects from analysis to launch Develop feedback loops from power-user traces and usage data 🎯 Requirements Strong fundamentals in ML, software eng, systems or stats Hands-on with LLMs, post-training, RL/RLHF/RLAIF, evals Ability to form hypotheses from transcripts, traces, failures Thrive in ambiguous problems with noisy signals Focus on developer/expert-user experience in real workflows Comfortable across research, product, infra, data, evals, safety