We're MindAlter Studio. We build AI characters — not branded content, not client work, not some agency's social strategy. Actual characters. With personalities, audiences, storylines, and fans who pay real money to follow them.

We run 12+ of these characters across TikTok, Instagram, and YouTube Shorts. They're generating $500K+/mo in revenue. And they're growing.

Here's the thing nobody outside our walls understands yet: the generation stack is the studio. Face consistency, model selection, the way pores catch light, the asymmetry of a real face, the difference between a frame that feels human and one that screams AI — that's not infrastructure. That's the product. There is no rendering farm to inherit. There is no playbook to follow. There's a frontier, a portfolio of characters, and what you build. That's it.

The Generative AI Lead doesn't support the content. The Generative AI Lead makes the content possible.

What you'd actually be doing.

Here's what your week looks like: You own generative AI at MindAlter — discovery, scaling, and maintenance. Full stop.

• Discovery. New models drop weekly. New techniques every month. You're the one who knows what's worth testing, what's hype, and what changes the game. You bring back the things that move the needle and ignore the rest.

• Scaling. We have 12+ characters today and we're growing. You decide how face consistency works across all of them. You decide which models become the standard. You decide where we invest in custom tooling and where off-the-shelf is enough. You build the stack that lets Character Owners ship without reinventing the wheel each time.

• Maintenance. When Kling outputs degrade, you know why. When a LoRA drifts, you fix it. When a character's face stops being recognizable across edits, you've already shipped the patch. The pipeline runs because you make it run.

And sometimes you'll look at what the creative team is asking for and say "that's not how this should be built — here's a better way." The team works with the tools, and you set the standard for what those tools need to be. If the stack isn't there, you don't duct-tape it — you architect it right.

Then you do it again. As the studio scales from 12 characters to 50.

Here's some of the stack that we expect you to know

You need to be genuinely, unreasonably good at this stack. Not "I've used it." Not "I'm familiar with it." Not "I took a course." We mean: when something breaks at 2am, you're the person who knows why. When a new model drops, you've already tested it by Friday. When someone shows you an AI video, you can reverse-engineer how they made it.

• Image generation models in production — Nano Banana 2 (Gemini 3.1 Flash), Flux 2, Imagen 4, Seedream, Ideogram. You've shipped with them, not just played in a playground. You know which model wins for character consistency, which for photorealism, which for typography, and you can defend the call.

• Video generation pipelines — Veo 3.1, Runway Gen-4.5, Sora 2, Kling 2.6, Luma Ray3, or whatever's current when you read this. You've worked at the frontier and know every failure mode by heart. Bonus if you've routed between them based on shot type.

• Face consistency at scale — LoRA training, IP-Adapter FaceID, PuLID Flux II, ControlNet. You know which tool solves which problem, why, and what the tradeoffs are. Ideally, you've trained your own LoRAs and know exactly why your last one failed.

• ComfyUI and node-based workflows — You build, debug, and optimize complex graphs in your sleep. Ideally, you've worked with PuLID + LoRA + ControlNet stacks for character consistency. Bonus if you've written custom nodes. Double bonus if you've open-sourced them.

• Voice and lip-sync — ElevenLabs v3, Fish Audio, HeyGen, Rask. You know which tool to reach for when a character needs to talk, and you've solved the cross-lingual voice preservation problem at least once.

• GPU infrastructure and cost control —You know exactly what a pipeline costs to run and how to bring that number down. You've made the call between local GPUs and API aggregators based on volume.

• Prompt engineering at the system level — Not "writing good prompts." Designing prompt architectures that produce consistent output across thousands of generations.

• Model evaluation and quality gates — You define what "good" means quantitatively and build the systems that enforce it. You know how to spot when a LoRA drifts before it ships.

• Creative-technical translation — You can take a creative brief and turn it into a technical spec — and push back when the brief is asking for something the stack can't deliver yet.

• Frontier awareness — You read the papers. You follow the right people. You test new releases the day they drop. You have strong opinions about what's worth adopting and what's noise.

$100-180k base salary. On top of that:

• Performance-based bonuses tied to studio growth. Your stack powers every character. When the studio scales, you scale.

• Full technical ownership. No legacy infrastructure. No "we've always done it this way." No committee deciding which models you get to use. You own the architecture.

• A portfolio of 12+ characters to build for. Variety, range, real production load. Not optimizing the same pipeline for 18 months until you stop learning.

• AI-native everything. We're not afraid of AI — we're built on it. You'll work with tools and techniques most engineers won't touch for another 3 years.

• Remote. Work from wherever your setup is. We care about what your stack outputs, not what timezone you're in.

Who we are, briefly.

We're building one of the most ambitious projects at the intersection of AI and internet culture. The characters exist. The systems work.

We're now scaling to build the most successful AI-native influencer brands on the internet. We move fast, ship constantly, and iterate based on data. No endless meetings. No bureaucracy. We value obsessive ownership, bias toward action, low ego, high accountability, and comfort with edgy, unconventional content.

If you naturally improve everything you touch and thrive in ambiguity — you'll fit right in. If you want a place where your technical decisions truly matter — and where you can see the direct impact of what you build across an entire portfolio — this is it.

-

We're currently hiring from the United States, Canada, South America, and Europe.

If we like what we see, we'll set up a call within 48 hours. No 6-round interview gauntlet. We'll know fast and so will you.

The future of influence is being built right now. Come be part of it. Show us what you've built. We're excited to meet you.