Feral Signal is hiring an AI Visual Workflow Specialist to help build, test, and refine advanced image and video generation workflows across ComfyUI, Forge, Automatic1111, and adjacent open-source generation tools.
This role is for someone who already lives in the weeds of generative media: checkpoints, LoRAs, ControlNet, IPAdapter, img2img, inpainting, upscaling, workflow graphs, prompt testing, video generation experiments, and the constant chaos of new nodes, models, forks, and pipelines.
You do not need to provide your own GPU power. Feral Signal will provide the compute required through our own enterprise in-house cloud instance.
We are looking for someone who can help us turn generative AI workflows into reliable internal systems: repeatable, documented, testable, and commercially useful.
This is not a generic "AI artist" role. We want someone who understands the tools deeply enough to troubleshoot broken workflows, compare model outputs, improve quality, document repeatable processes, and help us build visual generation systems that can be reused across product experiments, marketing assets, landing pages, research, and internal creative pipelines.
What you'll work on
You'll help Feral Signal with:
- ComfyUI workflow design, testing, and optimisation
- Forge / Automatic1111 pipeline testing
- Text-to-image workflows
- Image-to-image workflows
- Inpainting and outpainting workflows
- Image-to-video workflows
- Text-to-video workflows
- LoRA, checkpoint, sampler, scheduler, and model testing
- ControlNet / IPAdapter / reference-based generation workflows
- Upscaling, face/detail restoration, consistency workflows, and post-processing
- Building reusable workflow templates
- Documenting generation processes clearly enough for other operators to follow
- Comparing outputs across tools, models, and settings
- Creating internal SOPs, presets, prompt structures, and QA checklists
- Testing new open-source models, nodes, extensions, and generation stacks
- Supporting visual assets for landing pages, product experiments, brand systems, and growth campaigns
What we're looking for
You should be comfortable with at least some of:
- ComfyUI
- Forge
- Automatic1111
- Stable Diffusion / SDXL / newer open image models
- LoRAs, embeddings, checkpoints, VAEs, ControlNet, IPAdapter, etc.
- img2img, inpainting, reference workflows, and consistency workflows
- AI video generation tools and local/open workflows
- Prompt engineering for image/video generation
- Model testing and output comparison
- Workflow troubleshooting
- Writing clear notes, documentation, and repeatable instructions
You do not need a traditional degree or agency background. We care much more about taste, technical curiosity, practical execution, and whether you can make the tools behave.
The ideal person
The ideal person is a hobbyist-turned-power-user who already spends too much time testing models, breaking workflows, fixing nodes, comparing outputs, and figuring out why one pipeline produces magic while another produces plastic garbage.
You should have strong visual judgement, but this is not purely an "artist" role. We need someone who can think like an operator: test, document, improve, repeat.
You should be comfortable working remotely, communicating clearly, and turning messy creative experiments into systems other people can use.
Compensation
Monthly Pay: $2,465–$7,300
Structure: Freelance contractor
Location: Fully remote
Compute: Provided by Feral Signal
Hours: Flexible, based on agreed scope
Perks
This role includes access to $10,000+ in relevant tools, subscriptions, creator resources, and workflow infrastructure.
Depending on scope, this may include things like:
- Premium AI image/video generation tools
- ComfyUI / model workflow resources
- Civitai, model marketplace, or creator-support budgets
- Midjourney, Krea, Magnific, Leonardo, Freepik, RunDiffusion, or similar creative subscriptions
- Topaz Photo AI / Video AI
- Adobe Creative Cloud or equivalent editing tools
- Figma / design tooling
- Asset libraries, stock resources, plugin packs, and creator bundles
- Cloud storage for models, datasets, and outputs
- Replicate, Hugging Face, Stability, or other model/API credits where useful
- Hardware or workflow accessories where justified by the role, such as SSD storage, drawing tablet, calibration tools, or other production aids
Feral Signal provides the core compute needed for the work through our own enterprise cloud setup.
Where specific subscriptions, licenses, or resources are issued personally or agreed in writing as contractor perks, they may remain yours after the engagement ends. Company infrastructure, internal systems, private assets, and enterprise accounts remain Feral Signal property.
Category note
Some Feral Signal projects may involve emerging AI products, creator platforms, affiliate offers, online communities, and visual material that some individuals may find sensitive.
We expect professional judgement, discretion, and maturity. You do not need to work on anything that falls outside agreed boundaries, but applicants should be comfortable operating around category-sensitive digital products and creative workflows.
If you already know your way around ComfyUI graphs, Forge settings, broken nodes, weird model behaviour, and the difference between a lucky output and a repeatable workflow, we want to talk.