Lead Video Researcher

We are an elite, stealth-stage applied AI research lab in New York City backed by premier venture firms, pioneering AI researchers, and prominent tech founders. We are tackling one of the final frontiers in artificial intelligence: computational creativity.

Our mission is to build a foundational intelligence platform engineered for superhuman computational creativity. We believe this platform will spark a new creative renaissance, allowing anyone to direct, compose, and produce studio-grade work simply by conversing with AI.

We operate as a small, tight-knit, multidisciplinary team of world-class researchers, systems engineers, and acclaimed creative artists working together in our physical lab New York.

We are teaching machines to be genuinely creative, and video is where that challenge is hardest: motion, timing, composition, lighting, and story all have to hold together simultaneously.

Most generative video models today produce isolated 3-to-5-second clips fraught with flicker, drift, and zero narrative coherence. We are moving beyond novelty snippets to build autonomous creative video systems capable of orchestrating multi-shot narratives, preserving continuity across scenes, and translating raw human intent into production-grade films.

As our Video Lead, you will be the first scientist/engineer fully dedicated to the video modality, directly shaping our technical roadmap, model stack, and research agenda.

What You’ll Do

• Design & Build Agentic Video Systems: Architect LLM-powered agents that take high-level creative concepts through automated storyboard planning, multi-shot generation, camera orchestration, and iterative editing to deliver finished pieces.
• Embed Cinematic & Artistic Reasoning: Codify the judgment behind framing, lighting, pacing, camera movement, and cutting rhythms into mathematical loss functions, critique loops, and model evaluation verifiers.
• Orchestrate the Generative Stack: Compose generation models (DiT, video diffusion), editing frameworks, and vision-language models (VLMs) into low-latency production pipelines producing cohesive stories rather than disconnected clips.
• Pioneer Video Evaluation & Quality Gates: Design automated evaluation frameworks to verify agent decision-making throughout the workflow, alongside self-repair loops that detect and remediate artifacts, warping, or stylistic drift before output delivery.
• Ship Research to Production: Deploy and serve models at scale on dedicated multi-GPU clusters using distillation, quantization, progressive rendering, and Triton/CUDA acceleration with sub-second previews.
• Partner with World-Class Artists: Work daily with renowned filmmakers, cinematographers, and creative directors to translate aesthetic taste into training data and model controls.

Who You Are (Candidate Profile)

1. Technical Mastery & Applied Systems

• Proven Agentic or Generative Systems Experience: Hands-on experience building agentic or LLM-powered systems for video or media creation beyond simple prompt wrappers (e.g. multi-step planning, tool use, state tracking, and sequential asset delivery).
• Deep Familiarity with Video Architectures: Intimate knowledge of the modern generative video landscape—understanding current failure modes and trade-offs across Diffusion Transformers (DiT), video diffusion models, latent editing, and multimodal foundation models.
• Strong Software Fundamentals: Writes clean, modular, and maintainable production Python that team members can build upon.
• Experimental Rigor: A first-principles experimentalist with careful setups, measured claims, and conclusions grounded in empirical data.

2. Cinematic Taste & Intuition

• You understand video the way filmmakers do: why Kubrick centers a frame, why Tarkovsky holds a shot, or why a cut lands a beat early.
• You can articulate precisely why one generated shot works and another fails, and you possess the technical drive to translate that aesthetic intuition into model parameters.

3. Pedigree & Desired Background (Bonus Signals)

• Experience at a Tier-1 research lab or leading generative media company (e.g. Adobe Firefly, Runway, Snap, Meta FAIR, Google DeepMind, OpenAI, Stability AI).
• Peer-reviewed publications in premier computer vision, ML, or HCI venues (CVPR, ICCV, ECCV, SIGGRAPH, NeurIPS, ICML, ICLR, CHI, UIST).
• Practical background in VFX, 3D graphics, computational photography, or personal filmmaking.