What the work is
Join a leading AI lab's cutting-edge GenAI team to be at the core of the AI revolution, where your expertise fuels the development of the most advanced AI models.
Overview
A leading AI lab is building generative video models that aim to capture human performance and staging faithfully: acting realism, blocking, screen direction, eyelines, continuity and emotional nuance, carried into a new output. We're looking for experienced directors, assistant directors, animation supervisors and performance specialists to help define what a convincing, well staged performance looks like, and to judge where the models get it right or wrong. This is a full-time role of approximately 40 hours per week.
Key Responsibilities
- Set the quality bar: Define clear standards for acting realism, blocking, screen direction (the 180-degree rule and eyelines) and continuity, critique current evaluations, and help build a taxonomy of the ways generated performances fail.
- Curate benchmarks: Identify, generate and curate strong examples of excellent and genuinely hard dramatic performances and staging, and test early model capabilities against them.
- Critique model outputs: Review generated scenes the way you would in dailies, and write precise, actionable notes on performance, staging and continuity.
- Shape the reviewer pool and captions: Help define how human reviewers are vetted and calibrated, and review captioning guidelines so performance and directing nuance is described accurately.
- Collaborate with other experts to keep evaluations consistent and accurate.
Who gets hired
- 7+ years of professional experience in film, television, theater or animation.
- Credits on 2+ professional films.
- A background in at least one of: Film or Television Director or 1st Assistant Director; Animation Director or Character Animation Supervisor; Script Supervisor; or Performance Capture Director, Acting Teacher or Coach (film or theater).
- A meticulous eye for micro expressions, emotional beats, eyelines, continuity and staging.
- Hands-on use of generative image and video tools (e.g., Runway, Midjourney, Sora, Kling) to create your own samples for testing and comparison.
- Nice to have: reliable access to a 4K display for pixel-level review.
- Ability to engage reliably for approximately 40 hours/week during weekdays.
- Strong written communication and the ability to turn acting, blocking and directing detail into precise language for captions and evaluation templates.
The platform posted this on 22 Oct 2026.