Member of Technical Staff - Pretraining
Black Forest Labs is seeking a Member of Technical Staff to lead pretraining efforts for their multimodal foundation models, encompassing image, video, and audio. This role is central to the company's research initiatives, directly influencing products utilized by millions worldwide. The company, renowned for innovations like Latent Diffusion, Stable Diffusion, and FLUX, is headquartered in Freiburg, Germany, with an expanding presence in San Francisco.
The selected candidate will spearhead large-scale pretraining experiments, focusing on architecture design, objective functions, and scaling strategies. Responsibilities include developing and evaluating novel ideas across architectures, optimizers, and training algorithms, as well as contributing to the full stack—from low-level GPU and systems optimizations to high-level model design. The role also involves leading independent research projects and driving larger cross-team initiatives.
Ideal applicants will have led or co-owned pretraining for a foundation model that has been deployed to production or a major release. They should possess deep experience with large-scale distributed training, including multi-node runs at 500+ GPUs, and have a strong intuition for architecture and objective design. Proficiency in Python and PyTorch, along with familiarity with visual generative models, is essential.
The position offers a competitive base annual salary ranging from $180,000 to $300,000 USD. Black Forest Labs operates a hybrid work model, with expectations for in-person collaboration in Freiburg or San Francisco at least two days a week, or one full week every other week. The company covers reasonable travel costs to facilitate this arrangement.
Black Forest Labs values research excellence, open science, and technology that enhances human creativity. The team is grounded in four core values: obsession with quality, low ego, boldness in innovation, and kindness in interpersonal interactions. Joining this team offers the opportunity to work at the forefront of visual intelligence, contributing to technologies that shape the future of image and video creation.