SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Salary: GBP 150,000 - 300,000 / annual
Goodfire is a research company focused on AI interpretability—understanding how large language models work internally and using that understanding to make them safer, more useful, and more controllable. The company is backed by over $200M from top-tier VCs (B Capital, Menlo Ventures, Lightspeed) and founded by world-class interpretability researchers from OpenAI and DeepMind.
As a Research Scientist, you will conduct original research on interpretability mechanisms, develop novel techniques for understanding neural network internals, and help translate research into production tools. You'll work with a small, mission-driven team in the London office on three main areas: foundational research on how models represent information, high-risk/high-reward moonshot projects, and applied research that turns interpretability insights into real-world enterprise applications.
Key responsibilities include: conducting novel research in interpretability and adjacent fields; prototyping techniques to visualize and manipulate model internals; collaborating with engineers to productionize research; publishing and open-sourcing your work; and helping shape the company's research direction.
You'll need a PhD (or equivalent) in ML, computer science, or a quantitative field; deep familiarity with large language models; strong Python and PyTorch skills; and excellent communication abilities. Preferred: prior research leadership, open-source contributions, background in alignment or safe AI development, or startup/lab experience.
The company operates in-person 5 days per week from the London office, with one company-wide remote week per month. This is a full-time, permanent role with competitive salary, equity, and benefits.
