Member of Technical Staff - Safety
Posted Updated
Description
You will research and ship methods that make FLUX models safe by design, from training data to mid- and post-training, across images, video and text. You will develop safety SFT, preference optimization, RL and concept erasure methods, build data filtering at pretraining scale, make safety robust for open-weight releases, design evaluations and red-teaming, and work with pretraining and post-training teams.
Responsibilities
- Develop mid- and post-training safety methods for FLUX models such as safety SFT, preference optimization, RL with safety rewards and concept erasure
- Build safety data filtering classifiers and pipelines at pretraining scale
- Make safety robust to adversarial prompts and downstream fine-tuning for open-weight releases
- Extend safety to prompts, captions and text encoders
- Design evaluations and red-teaming measuring harmful generations alongside quality and prompt following
- Work with pretraining and post-training teams and publish where it makes sense
Requirements
- Hands-on research experience in visual safety of text-to-image, text-to-video or vision-language models
- Proven experience in large-scale safety data filtering or safety mid-/post-training
- Deep familiarity with concept erasure methods and their failure modes
- Track record of shipping: top-venue publications paired with production impact
- Experience with distributed training at multi-node scale
- Strong PyTorch fluency
- Experience working with coding agents and multi-agent workflows
Benefits
- Equity
- Coverage of reasonable travel costs to the office
Skills
Sign in to read and join the discussion.