Project Touchstone
I'm working on the Handshake project as a Manager 1, where I write prompts and rubrics that train AI models to complete creative and technical tasks across disciplines like video editing, 3D graphics, audio, and game development. I write prompts that read like real project briefs — the way a manager would brief a skilled employee — focused on what the final deliverable should achieve rather than micro-specifying every detail. Then I build binary rubrics that validate whether the AI's output meets the critical functional requirements without over-decomposing the work. I review golden examples, align with pod leads, and ensure every prompt-rubric pair balances functional precision with enough room for reasonable creative interpretation. The goal is to make AI outputs production-ready by teaching the models what "good" looks like through well-structured human evaluation.