Lead Product Content Engineering - Creative Evals

Product Content Engineering is a horizontal function supporting initiatives across Instagram and Meta AI, and is seeking a content engineer to lead media generation evaluations. We partner closely with product, research and modeling teams to set standards of quality and define what “good” looks like for generative image and video experiences. This position is focused on model training for image and video generation capabilities. You will own the human evaluation layer that tells our models what quality means, designing the rubrics, running the eval rounds, and turning subjective creative judgment into reproducible, measurable signals that directly shapes model behavior and launch decisions. The ideal candidate has built video and image evaluation rubrics from scratch and has hands-on video production expertise — they can articulate why a generated clip fails on motion coherence, lighting continuity, subject consistency, pacing, framing or audio sync, and can write that judgment down so that dozens of raters apply it the same way. They are fluent in text-to-video, image-to-video and video-to-video generation, and in the failure modes specific to each. You will operate as the quality authority across multiple modeling workstreams: setting the taxonomy, arbitrating disagreement, calibrating rater pools, and translating eval results into clear guidance for researchers and product leaders. You will also drive where human evaluation should give way to automated or LLM-as-a-judge measurement, and prove out that transition with data.