Generative Media Engineer (AI Video & AI Image Models)

Stellenbeschreibung:

At neural frames, we’re building a first‑of‑its‑kind interface that helps musicians and AI artists create AI music videos. We’re a very fast‑growing, profitable AI startup based in Berlin, and we live at the frontier of generative video, music, and visual storytelling.

We’re looking for a Generative Media Engineer who loves AI image & AI video models and wants to push the limits of what’s possible in AI music video generation. You’ll work directly with the founder and core team to design, fine‑tune, and productionise the next generation of AI video models and workflows.

If you’re excited about ComfyUI graphs, diffusion models, video pipelines, and agentic workflows - and you want your work to shape how thousands of artists create visuals for years to come - this role is for you.

Tasks

Manipulate and fine‑tune AI video models (and image models where needed) to power our music video and video generation features.
Curate, build, and maintain datasets for training and fine‑tuning: collecting, cleaning, tagging and evaluating data quality with a strong eye for visual and stylistic consistency.
Prototype and productionise agentic workflows for generating full AI music videos end‑to‑end (storyboarding, shot planning, style control, lip sync, motion consistency, etc.).
Build and optimise ComfyUI (or similar) pipelines for internal experimentation and, where useful, user‑facing tools.
Stay on top of the open‑source generative media ecosystem (ComfyUI nodes, new diffusion/video models, control techniques, schedulers, etc.) and bring the best ideas into our product.
Collaborate closely with our engineers to ship models into production : packaging, inference optimisation, GPU performance, monitoring, and safety guardrails.
Define and improve evaluation metrics & tooling (quantitative and qualitative) for visual quality, temporal coherence, and user‑perceived “wow”.
Work with the founder and product team to explore new creative capabilities (e.g. style transfer, camera motion, character consistency, audioreactive visuals) and turn them into features that artists actually use.

Requirements

Strong Python skills and experience working with deep learning frameworks (ideally PyTorch).
Hands‑on experience with AI image and/or video generation : diffusion models, video diffusion, control networks, or similar architectures.
Practical experience with ComfyUI or comparable node/graph‑based generative tooling (e.g. custom nodes, complex graphs, automation).
Experience in at least one of the following:
- Fine‑tuning or training diffusion or video models
- Building custom inference pipelines for generative image/video
- Designing automated/agentic workflows around generative models
Comfort working with GPU‑based workloads (e.g. VRAM constraints, batching, mixed precision, model pruning/quantisation, etc.).
You’re excited about the creative side of the work: music videos, visual storytelling, motion, style, and aesthetics matter to you.
You enjoy a fast‑paced startup environment : you iterate quickly, own projects end‑to‑end, and are okay with some chaos as long as we’re moving forward.
Location: You are based in Berlin or willing to relocate.

Benefits

Shape the frontier of AI music videos : Your models and workflows will directly define what thousands of artists can create with Neural Frames.
High impact, low bureaucracy : Small, focused team. You ship things that matter, quickly.
Work with people who care about art & tech : We’re an arts‑driven, creator‑obsessed team that genuinely cares about the visuals, not just the metrics.
Profitable, fast‑growing AI startup : Rare combination of real revenue, real users, and room to experiment.
Autonomy & ownership : Drive your own roadmap inside a clear product context. If you have a strong idea, we’ll probably try it.
Hybrid work in Berlin : Enjoy a mix of in‑person collaboration in our Berlin office and flexible time working remotely.