Research Scientist
Goodfire is a San Francisco AI interpretability research company building Silico, an agent and infrastructure for understanding, debugging, monitoring, and controlling AI models.
Funding history
About Goodfire
Goodfire is a public benefit corporation and AI interpretability research lab. Its current product, Silico, turns research questions into inspectable experiments and reports, supporting model analysis, debugging, guardrails, and interpretability research.
Skills
Candidate Availability
Required and preferred rules are kept separate and reflect the wording in the original posting.
About the Role
You will conduct original interpretability research, prototype methods for visualizing and manipulating internal model structures, and collaborate with engineering to turn research into production-ready tools. You will share results through publications, demos, and open-source contributions and help shape research direction, with a focus on mechanistic understanding of neural-network algorithms.
Requirements
- PhD or equivalent experience in ML, computer science, or a quantitative science.
- Deep familiarity with large models.
- Fluency in Python and ML frameworks such as PyTorch.
- Strong writing and communication skills.
- Experience leading research or contributing to open-source codebases.
- Familiarity with interpretability, alignment, or safe model development.
- Experience in startup or fast-paced lab environments.
Responsibilities
- Conduct original research in interpretability and related fields.
- Prototype techniques to visualize and manipulate internal model structures.
- Collaborate with engineering to turn research into production-ready tools.
- Share work through publications, demos, and open-source contributions.
- Help define and evolve research direction.
Benefits
- Equity
- Competitive benefits
- One company-wide remote week per month