Overview The base model provides capability; post-training turns that capability into a product people want to use. This is a technical, hands-on PM role embedded with applied researchers and ML engineers conducting RLHF, fine-tuning, and preference tuning, translating model outputs into concrete training priorities. You will ensure the end-user voice is reflected in a process driven by metrics and iteration, working with the research team to shape post-training work. Responsibilities - Embed in the post-training loop, working day-to-day with research and applied ML teams during fine-tuning and RLHF cycles, reviewing model outputs and giving structured feedback on quality, tone, and behavior - Define what 'good' means for Copilot models: set priorities for capability, personality, and safety tradeoffs across surfaces (Microsoft 365 Copilot, Copilot Studio agents, Windows) - Run RLHF/preference data campaigns: partner with data and labeling teams to scope what human preference data gets collected and why - Build and maintain behavior evals: translate qualitative judgments into evaluation sets tracked release over release - Own the feedback loop: turn user research, enterprise customer feedback, and production incident learnings into concrete post-training priorities - Make tuning tradeoffs explicit: helpfulness vs. caution, consistency vs. personality, latency/cost vs. quality; drive alignment across research, safety, and product leadership on where the line sits - Represent Responsible AI and enterprise requirements in training

Overview The base model provides capability; post-training turns that capability into a product people want to use. This is a technical, hands-on PM role embedded with applied researchers and ML engineers conducting RLHF, fine-tuning, and preference tuning, translating model outputs into concrete training priorities. You will ensure the end-user voice is reflected in a process driven by metrics and iteration, working with the research team to shape post-training work. Responsibilities - Embed in the post-training loop, working day-to-day with research and applied ML teams during fine-tuning and RLHF cycles, reviewing model outputs and giving structured feedback on quality, tone, and behavior - Define what 'good' means for Copilot models: set priorities for capability, personality, and safety tradeoffs across surfaces (Microsoft 365 Copilot, Copilot Studio agents, Windows) - Run RLHF/preference data campaigns: partner with data and labeling teams to scope what human preference data gets collected and why - Build and maintain behavior evals: translate qualitative judgments into evaluation sets tracked release over release - Own the feedback loop: turn user research, enterprise customer feedback, and production incident learnings into concrete post-training priorities - Make tuning tradeoffs explicit: helpfulness vs. caution, consistency vs. personality, latency/cost vs. quality; drive alignment across research, safety, and product leadership on where the line sits - Represent Responsible AI and enterprise requirements in training