3 months ago
Base Salary
$250k - $500k/yr
Responsibilities
- Create diverse evaluations spanning a range of difficulty levels.
- Identify naturally occurring interesting and undesirable behaviors in open-source models.
- Develop novel architectures and training objectives for oversight assistants.
- Scale training and inference pipelines to support models up to 1T scale.
- Conduct reliable experiments and communicate findings transparently.
- Collaborate with a highly interactive research and engineering team.
Requirements
- Experience fine-tuning language models.
- Experience designing new model architectures.
- Experience creating evaluations.
- Strong programming ability, including balancing prototyping speed with maintainability.
- Strong experimental design and epistemic self-awareness.
- Ability to generate original ideas and unblock research progress.
- Curiosity about how machine-learning systems work.
- Strong communication skills and openness to feedback.
Benefits
- Benefits are provided.
- The role is based in San Francisco and involves working together in person.
- The employer is open to sponsoring international visas.
Categories
AI ResearchML Engineering
About Transluce
Transluce is an independent nonprofit research lab in San Francisco, founded in 2024, that builds open-source tools to analyze and evaluate advanced AI systems. It develops automated assessments of behaviors like honesty, misreporting, and evaluation awareness on open-weight models, then works with frontier AI labs and governments to apply vetted procedures. The lab focuses on public-interest AI transparency, reliability, and scalable oversight.