Forward Deployed Engineer (Generalist)

What if the gap between groundbreaking research and real-world application wasn't a chasm, but a bridge you help build?

We're a ~50-person team whose models—Latent Diffusion, Stable Diffusion, FLUX.1—have been downloaded 400M+ times. More than Google and Microsoft combined. But downloads don't tell the whole story. Between a research breakthrough and a production system lies territory that's equal parts technical puzzle and human collaboration. That's where you come in.

What You'll Pioneer

You won't just integrate our models—you'll discover what's possible when cutting-edge generative AI meets real-world constraints. You'll work directly with customers who are trying to do things that have never been done before, which means you'll encounter problems that don't yet have Stack Overflow answers.

You'll be the person who:

• Designs deep product integrations for customers navigating everything from model hosting and finetuning to deployment and inference optimization
• Builds intuitive interfaces that make sophisticated diffusion models feel approachable—whether that's on top of existing integrations or entirely new ones
• Helps customers deploy FLUX models in their environments (both BFL-hosted and on-premise), optimizing for the delicate balance of latency and output quality
• Sits across the table from customers to understand not just what they're asking for, but what they actually need—then architects solutions accordingly
• Explores uncharted territory: What industries haven't yet discovered what's possible with generative visual AI?

Questions We're Wrestling With

• How do we make FLUX as easy to deploy as it is powerful to use?
• What does "optimized inference" actually mean when every customer's constraints are different?
• Where's the line between what should be hosted and what should run on-premise—and how do we help customers make that call?
• What emerging use cases are we not seeing yet because the tooling doesn't make them obvious?
• How do you explain transformer architectures and flow matching to someone who just needs their creative pipeline to work?

These aren't rhetorical. We're figuring this out as we go, at the frontier.

Who Thrives Here

You're energized by the space between "research breakthrough" and "production system"—that messy, fascinating territory where theory meets reality. You've shipped software that real people (or real companies) depend on. You can talk about API design with backend engineers in the morning and explain latency-quality tradeoffs to non-technical stakeholders in the afternoon.

You likely have:

• A track record of working directly with customers on generative AI deployment, iterating on solutions in real-time rather than tossing documentation over a wall
• Comfort across the full stack—from model hosting and backend architecture to building interfaces that people actually want to use
• Strong Python skills and an intuitive understanding of API integrations (because prototypes and demos matter)
• The rare ability to translate sophisticated technical concepts into language that makes sense for whoever's in the room

We'd be especially excited if you:

• Have hands-on experience with diffusion models, flow matching, or related finetuning and distillation techniques
• Know your way around the FLUX ecosystem—ComfyUI, common training frameworks, the tools builders are actually using
• Have optimized inference for transformer-based models and lived through the tradeoffs
• Have architected solutions in complex enterprise environments where "just use the cloud" isn't always the answer
• Contribute to open-source projects, particularly around diffusion models

What We're Building Toward

We're not just deploying models. We're learning what it takes to make frontier research useful in the real world—with all its messy constraints, competing priorities, and unexpected use cases. If that sounds more interesting than having all the answers, we should talk.

Base Annual Salary: $180,000–$300,000 USD

We're based in Europe and value depth over noise, collaboration over hero culture, and honest technical conversations over hype. Our models have been downloaded hundreds of millions of times, but we're still a ~50-person team learning what's possible at the edge of generative AI.