Job Description
Start Date: Q1 2027
Base Salary: $120,000–$140,000
Location: On-Site (100% In-Office)
Hours: Monday – Friday | 8:00 AM – 6:00 PM
About the Role
Subject produces thousands of educational videos each year. Thousands. This role will be crucial in supporting that goal while keeping quality at the forefront of everything. We need locked characters, consistent worlds, no hallucinations and voiceovers that pronounce words correctly, and we need to do it at scale, automated. We’re looking for an AI Pipeline producer who does exactly that and will refine, run, and improve our existing process.
This is a hybrid role that barely exists in the market yet, which is why we're being specific. You are a ComfyUI expert. Not "I've opened it and connected some nodes." You build custom workflows, you understand what's happening under the hood, and you can debug a broken graph at 11pm because #deadlines. You also come from real production. You know the production process end to end. You know why a scene is blocked the way it is, why lighting continuity matters, why a cut lands or doesn't. We’re looking for a high-level production understanding.
You'll work at the center of our AI production operation, building pipelines in ComfyUI and Python, managing generation workflows with access to engines from Seedance, Luma, Higgsfield, Kling, HeyGen, etc and produce finished, polished video that stands next to our live-action work.
Why Subject
Most AI video jobs are marketing gigs with a Runway subscription. This one is a production role at a company where the AI pipeline ships real curriculum to real students at a scale nobody else is attempting. If you've been waiting for a place where your ComfyUI obsession and your production background finally belong in the same job, this is it.
What You'll Do
Build, maintain, and improve video generation pipelines in ComfyUI, including custom nodes and Python tooling
Produce AI video content end to end: prompt systems, character consistency, style bibles, negative prompt strategy, and QC
Manage high-volume generation workflows where the unit of work is hundreds of videos, not one
Bring production discipline to AI output: shot logic, continuity, pacing, and coverage that a director would recognize
Create and integrate motion graphics that elevate generated footage into finished lessons
Debug the weird stuff: duration drift, model quirks, API failures, render bottlenecks
Evaluate new models and tools as they drop, and decide fast whether they earn a place in the pipeline
Work alongside our physical production team so AI and live-action content share one visual standard
Dabble in interactive and partner with our product design team to deliver assets
What We're Looking For
3 to 5 years of professional experience across production, post-production, or technical creative work
Present 5 Days a week in our studio 8am-6pm
Expert-level ComfyUI skills: custom workflows, custom nodes, and real fluency with the underlying models and samplers
Working knowledge of Python, enough to script, automate, and extend pipelines
Genuine physical production experience. You've been on sets, you understand how shoots actually work, and it shows in your generated output
Motion graphics experience (After Effects or comparable)
Hands-on experience with current AI video tools: Seedance, Luma, Higgsfield, Runway, Kling, HeyGen, or similar
Startup or fast-moving studio background. Ambiguity and shifting priorities don't rattle you
A portfolio that proves it. We want to see AI work with taste!
Nice to Have
Experience with lip sync and avatar pipelines (OmniHuman, HeyGen, Arcads, or similar)
ffmpeg fluency and comfort with command-line media workflows
Voice generation experience (ElevenLabs or similar)
LoRA training or model fine-tuning experience
Educational or narrative content background