See how candidates think and approach the work this role demands, before the phone screen. We'll build a video challenge from this posting, and you can edit or share it before it goes live.
Black Forest Labs is building a dedicated product-research function that turns evals, leaderboards, and customer usage into ranked model priorities, signaling a shift from research-driven model building toward product-led, market-measured image and video model development.
We’re the team building the foundational layer for visual intelligence: multimodal models trained jointly on images, video, audio, and action to see, generate, and understand the world.
Our founding team pioneered latent diffusion and Stable Diffusion, the breakthroughs that made generative AI accessible to millions. With FLUX, we're pushing into generative media at scale, machines that act in the physical world, and emerging applications in computer use.
Our FLUX models top public benchmarks and run in production in some of the world's biggest platforms. We're a small, fast growing research lab competing with the largest companies in the world. From our labs in Freiburg to San Francisco, we're building what comes next.
Our model research moves fast. What's missing is someone accountable for one question: what should our models get better at next, and how will we know when they have?
You’ll lead Product Research, the team between our researchers and the market. You’ll take raw signal from evals, leaderboards, customers and usage data, turn it into a clear measure of where our models stand, and decide which gaps matter most. Your squad of Applied Researchers runs the evals and closes those gaps through post-training. You’ll work closely with Research, Backend, Product and GTM.
Evaluations. Own the product-facing eval taxonomy, eval strategy, and eval runs.
Priorities. Maintain a ranked list of what our models should get better at, built from evals, leaderboards, customer feedback and usage data, and make sure it shapes every research cycle.
Applied post-training. Set the priorities for the squad's Applied Researchers, who finetune and run RL to improve our core models and to build variants for specific customers and products. You decide what's worth building, define what success looks like, and check that results hold up with no regressions.
Data. Define the data we need to measure and improve our models.
Model releases. Co-lead new model releases. You own the quality go/no-go and the capability claims.
PM experience (product or research) working directly on generative models. You've shaped the model itself rather than building features on top of it.
Real technical depth. You can read an eval design and spot what's wrong with it, and you understand finetuning and RL well enough to scope and challenge the work. Researchers take you seriously because of what you know.
Strong judgment on trade-offs made with incomplete signal, and the drive to own prioritization end to end.
Comfort in an ambiguous, fast-changing research environment, where the right priority shifts as the model and harness evolve.
You've done at least one of these:
Defined what a generative model should get better at, and got a research team to act on it.
Turned a messy backlog of model weaknesses into ranked priorities that survived contact with a research org. Sources might include bug reports, leaderboard gaps, and customer complaints.
Owned evaluation for a model release, including the ship call or the capability claims.
Scoped and shipped a finetuned model or variant for a specific customer or product need.
Diagnosed a quality regression or plateau and traced it to a data, training, or eval-design cause.
Nice to have: hands-on eval or ML engineering experience, and experience with diffusion or video models.
We’re a distributed team with real offices that people actually use. This role is based in Freiburg, and you’ll join us in the office at least 2 days a week, or one full week every other week.
Everything we do is grounded in four values:
Obsessed. We are a frontier research lab. The science has to be right, the understanding deep, the product beautiful.
Low Ego. The work speaks. The best idea wins, no matter who said it. Credit is shared. Nobody is above any task.
Bold. We take the ambitious bet. We ship, we do not wait for conditions to be perfect.
Kind. People over politics. We treat each other with genuine warmth. Agency without empathy creates chaos.
If this sounds like work you’d enjoy, we’d love to hear from you.
Compensation : 172K-220K base + equity, depending on profile
Note: We may use AI-assisted tools to help organize applications. All hiring decisions are always made by our team.
blackforestlabs:9f76d62d-3487-459f-bceb-2295f47366f1See how we measure for definitions, or our corrections log for known issues. Found something wrong? Flag a correction.
black-forest-labsThis posting hasn't been probed by our closure verifier yet. Stream C runs on a rolling schedule against postings approaching the close-decision threshold.