United States onlyWe ship real commercials for real brands on real media. We’re looking for someone who lives at the prompt terminal—turning text into moving images, sound, and finished ad assets. You’ll help us define the craft: build prompt defaults and behind-the-scenes “governors” for our live product, wire up node-based workflows, and constantly test the latest models so our creative stays sharp, fast, and on-brand.
Please include the following or your application will not be screened:
Please include the following or your application will not be screened:
- Reel (45–120s): your best AI-generated sequences or composites (link to YouTube, Frame, or similar required)
- Three prompts: paste exact prompts + key params (model, duration/AR/seed or approach) with links or stills
- One BTS artifact: a screenshot or brief description of a node graph/flow you built (can redact client names)
- 2–3 lines on a real constraint you solved (e.g., legibility, motion continuity, brand rules)
What You'll Do
- Generate at scale: Create short, cinematic sequences and ad components daily (shots, transitions, graphics, VO/music selects) for use in finished spots.
- Establish prompt systems: Build reusable prompt defaults (style, lensing, motion cues, timing), BTS governors (guardrails for tone/variance/safety), and shot-builder macros.
- Wire node-based flows: Stand up and iterate node graphs (e.g., ComfyUI/Houdini-style thinking, or similar) for repeatable pipelines—ingest → transform → QC → export.
- Survey models & update the stack: Evaluate new releases (e.g., Sora/Veo-class video, image, audio/VO) for cost/latency/quality; write short notes and example prompts; propose upgrades.
- Quality control: Stress-test outputs against brand/spec requirements (aspect, cadence, legibility, rights/safety); tune prompts to reduce artifacts and improve determinism.
- Document & share: Keep a living prompt library with “before/after” examples, parameters, and quick BTS context so others can reproduce your results.
What You'll Bring
- A reel (even short) that proves you can generate compelling moving images and/or composites using gen-AI tools—we screen reels first
- Fluency with at least two modalities (video/image/audio/VO) and comfort jumping between them
- Strong prompt craft: you think in shots, beats, tone, motion, and constraints (seed, camera, duration, typography, brand guardrails)
- Node-based problem solving (any graph tool or equivalent mindset)
- Clear written notes: you annotate your work so others can ship it
- Huge plus: shipped work in story/creative/ad contexts (agency, brand, film/TV, social, motion design, editorial).
Tools you may touch
Leading video models (e.g., Sora/Veo or similar), image (e.g., Midjourney/SD or similar), audio/music (e.g., Udio/Suno), VO/TTS, NLE/motion tools, node composers (e.g., ComfyUI/Houdini/Nuke-style thinking). We don’t require a specific stack, results matter most.
Benefits
Remote-friendly, no travel. We ship, learn, and improve the system together. This is a contract-to-FullTime conversion (over 3-6 months). We'll agree on a day-rate and go from there. Benefits + growth path shared early in process.
CA and US only
