About Black Forest Labs

We're the team behind Latent Diffusion, Stable Diffusion, and FLUX — foundational technologies that changed how the world creates images and video. Our models power the tools used by millions of creators, developers, and businesses worldwide, and FLUX is among the most advanced generative systems in the world.

Headquartered in Freiburg, Germany with a growing presence in San Francisco, we're scaling fast while staying true to what makes us different: research excellence, open science, and building technology that expands human creativity.

Why This Role

Vision-language models are becoming foundational to how people interact with generative AI — but most VLM research happens in isolation from the generation stack. At Black Forest Labs, we're integrating VLMs directly into FLUX in ways that make our models more powerful, more controllable, and more aligned with what creators actually want.

This role is about pioneering that integration. You won't be applying off-the-shelf VLMs — you'll develop novel approaches, innovate on architectures, and answer questions that haven't been solved yet: how vision and language representations inform each other, how multimodal understanding improves generation quality, and how to make these capabilities deployable at scale without compromising what makes FLUX exceptional.

This is a Staff / Senior IC role. We're looking for someone who has pretrained or significantly advanced a VLM, not just fine-tuned one.

What You'll Work On

• Lead development and training of state-of-the-art multimodal vision-language models within the FLUX stack — innovating on architectures, not just applying existing ones
• Design fine-tuning strategies that adapt VLMs to specialized creative use cases (captioning, editing instructions, prompt enhancement) that general-purpose models can't handle
• Research integrations between VLM/LLM capabilities and our diffusion and flow pipelines — finding creative ways to improve generation quality and controllability without computational bottlenecks
• Evaluate emerging multimodal architectures, translating the best of recent research into practical improvements

What We're Looking For

• You've pretrained or significantly advanced a VLM (not just SFT'd or LoRA'd one) that was deployed in a production system or released publicly
• Strong publication record or unambiguous production track record showing you push the frontier on multimodal architectures
• Deep understanding of how vision and language representations interact: tokenization, alignment, grounding, cross-modal attention, and the failure modes of each
• Experience with distributed training at multi-node scale
• Comfortable at the research/production boundary — you care whether the work ships and generalizes, not just whether it reads well
• Experience with diffusion or flow-based generative models is a strong plus — especially if you've thought about how autoregressive and diffusion paradigms can compose

How We Work Together

We’re a distributed team with real offices that people actually use. Depending on your role, you’ll either join us in Freiburg or SF at least 2 days a week (or one full week every other week), or work remotely with a monthly in-person week to stay connected. We’ll cover reasonable travel costs to make this possible. We think in-person time matters, and we’ve structured things to make it accessible to all. We’ll discuss what this will look like for the role during our interview process.

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.

About Black Forest Labs

We're the team behind Latent Diffusion, Stable Diffusion, and FLUX — foundational technologies that changed how the world creates images and video. Our models power the tools used by millions of creators, developers, and businesses worldwide, and FLUX is among the most advanced generative systems in the world.

Headquartered in Freiburg, Germany with a growing presence in San Francisco, we're scaling fast while staying true to what makes us different: research excellence, open science, and building technology that expands human creativity.

Why This Role

Post-training is where a foundation model becomes a product. In this role, you'll own the post-training pipeline for our multimodal models end to end — from data strategy and reward modeling to preference optimization, distillation, and safety tuning — across image, editing, and video. You'll drive measurable gains in model quality, build the infrastructure that lets the whole research team iterate fast, and push the state of the art in what it means to align a generative model to human intent.

This is a Staff / Senior IC role. We're looking for someone who has shipped post-training for a frontier model before and wants to do it again.

What You'll Work On

• Own the full post-training pipeline end to end — from data curation and reward modeling through fine-tuning, preference optimization, distillation, safety tuning, evaluation, and deployment
• Advance techniques across the post-training stack: SFT, RLHF, RLAIF, DPO, preference learning, and reward modeling to align models with human intent and aesthetic judgment
• Work across modalities: text-to-image, image editing, multi-reference, and video post-training
• Build personalization and customization capabilities that let users adapt our models to their own creative style
• Design and maintain high-throughput fine-tuning and evaluation infrastructure to support rapid iteration across the research team
• Identify quality and alignment gaps through rigorous evaluation, then close them through targeted research and engineering

What We're Looking For

• You've owned post-training for a frontier generative model through release (SFT, preference optimization (DPO or RLHF), distillation, safety tuning) with measurable quality wins on human prefs or standard benchmarks
• Deep experience across the post-training stack, not just one slice: reward modeling, preference learning, RLHF/RLAIF, and personalization
• Comfortable working across modalities: text-to-image, image editing, multi-reference, and ideally video
• Strong PyTorch fluency; you write research code that others can build on
• Experience with distillation (LADD, DMD, consistency models, or similar) or with building high-throughput eval pipelines is a strong plus
• Bias toward shipping: measurable model-quality improvements that reach users, not just papers

How We Work Together

We’re a distributed team with real offices that people actually use. Depending on your role, you’ll either join us in Freiburg or SF at least 2 days a week (or one full week every other week), or work remotely with a monthly in-person week to stay connected. We’ll cover reasonable travel costs to make this possible. We think in-person time matters, and we’ve structured things to make it accessible to all. We’ll discuss what this will look like for the role during our interview process.

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.

About Black Forest Labs

We're the team behind Latent Diffusion, Stable Diffusion, and FLUX — foundational technologies that changed how the world creates images and video. Our models power the tools used by millions of creators, developers, and businesses worldwide, and FLUX is among the most advanced generative systems in the world.

Headquartered in Freiburg, Germany with a growing presence in San Francisco, we're scaling fast while staying true to what makes us different: research excellence, open science, and building technology that expands human creativity.

Why This Role

We're building the foundation models that power the next wave of visual intelligence — and pretraining is where that work begins. This role sits at the center of our research effort. You'll shape training objectives, architectures, data strategies, and systems behind our joint image, video, and audio foundation models, with a direct line from your research to products used by millions.

This is a Staff / Senior IC role. We're looking for someone who has already led pretraining at the frontier and wants to do it again.

What You'll Work On

• Lead large-scale pretraining experiments for our multimodal (image, video, audio) foundation models (architecture, objective functions, scaling strategies)
• Develop and evaluate novel ideas across architecture, optimizers, and training algorithms.
• Contribute across the full stack: low-level GPU and systems optimizations, research code, and high-level model design
• Lead focused research projects independently and drive larger cross-team initiatives

What We're Looking For

• You've led or co-owned pretraining for a foundation model (image, video, LLM, or multimodal) that shipped to production or a major release
• Own architectural calls that move the model: attention patterns, modulation schemes, loss formulations, tokenization strategies
• Deep experience with large-scale distributed training: FSDP/TP/PP, multi-node runs at 500+ GPUs, debugging loss spikes, NaNs, throughput regressions, and silent correctness issues at scale
• Strong intuition for architecture and objective design — you've made calls on attention patterns, modulation schemes, or loss formulations that moved a real model
• Track record of shipping: top-venue publications (NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV) paired with production impact, or unambiguous production wins at a frontier lab
• Deep Python and PyTorch proficiency; comfortable reading and modifying low-level training code
• Familiarity with visual generative models is a must

How We Work Together

We’re a distributed team with real offices that people actually use. Depending on your role, you’ll either join us in Freiburg or SF at least 2 days a week (or one full week every other week), or work remotely with a monthly in-person week to stay connected. We’ll cover reasonable travel costs to make this possible. We think in-person time matters, and we’ve structured things to make it accessible to all. We’ll discuss what this will look like for the role during our interview process.

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.

About Black Forest Labs

We’re the team behind Latent Diffusion, Stable Diffusion, and FLUX—foundational technologies that changed how the world creates images and video. We’re creating the generative models that power how people make images and video—tools used by millions of creators, developers, and businesses worldwide. Our FLUX models are among the most advanced in the world, and we’re just getting started.

Headquartered in Freiburg, Germany with a growing presence in San Francisco, we’re scaling fast while staying true to what makes us different: research excellence, open science, and building technology that expands human creativity.

Why This Role

You'll train large-scale diffusion models for image and video generation, exploring new approaches while maintaining the rigor that helps us distinguish meaningful progress from incremental tweaks. This isn't about following established recipes—it's about running the experiments that clarify which architectural choices matter and which are less impactful.

What You’ll Work On

• Trains large-scale diffusion transformer models for image and video data, working at the scale where intuitions break and empirical evidence matters
• Rigorously ablates design choices—running experiments that isolate variables, control for confounds, and produce insights you can actually trust—then communicating those results to shape our research direction
• Reasons about the speed-quality tradeoffs of neural network architectures in production settings where both constraints matter simultaneously
• Fine-tunes diffusion models for specialized applications like image and video upscalers, inpainting/outpainting models, and other tasks where general-purpose models aren't enough

What We’re Looking For

You've trained large-scale diffusion models and developed strong intuitions about what matters. You know that at research scale, every design choice has tradeoffs, and the only way to know which ones are worth making is through careful ablation. You're comfortable debugging distributed training issues and presenting research findings to the team.

You likely have:

• Hands-on experience training large-scale diffusion models for image and video data, with practical knowledge of common failure modes and what matters most in training
• Experience fine-tuning diffusion models for specialized applications—upscalers, inpainting, outpainting, or other tasks where understanding the domain matters as much as understanding the architecture
• Deep understanding of how to effectively evaluate image and video generative models—knowing which metrics correlate with quality and which are just convenient proxies
• Strong proficiency in PyTorch, transformer architectures, and the full ecosystem of modern deep learning
• Solid understanding of distributed training techniques—FSDP, low precision training, model parallelism—because our models don't fit on one GPU and training decisions impact research outcomes

We'd be especially excited if you:

• Have experience writing forward and backward Triton kernels and ensuring their correctness while considering floating point errors
• Bring proficiency with profiling, debugging, and optimizing single and multi-GPU operations using tools like Nsight or stack trace viewers
• Know the performance characteristics of different architectural choices at scale
• Have published research that contributed to how people think about generative models

How We Work Together

We’re a distributed team with real offices that people actually use. Depending on your role, you’ll either join us in Freiburg or SF at least 2 days a week (or one full week every other week), or work remotely with a monthly in-person week to stay connected. We’ll cover reasonable travel costs to make this possible. We think in-person time matters, and we’ve structured things to make it accessible to all. We’ll discuss what this will look like for the role during our interview process.

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.