Research Scientist/Engineer — Evaluations at Apollo Research
Apollo Research is seeking a Research Scientist/Engineer for its pre-deployment team in London and San Francisco. The role involves designing automated pipelines for assessing AI model alignment and scheming. The ideal candidate has strong software engineering skills, experience with data analysis,
Apollo Research is hiring a Research Scientist/Engineer to work on Training-Run Assessments, evaluating frontier AI models for scheming and misalignment. The role is full-time and on-site, with thousands of runs across hundreds of distinct environments approximately every two weeks. The ideal candidate has 5+ years of experience in software engineering and data analysis.
Overview
The Research Scientist/Engineer will be part of a pre-deployment team working on evaluating and red-teaming AI models, with opportunities to interact with new models before they are released. The role involves collaborating with frontier labs like OpenAI, Anthropic, and Google DeepMind.
- Design and build automated pipelines for assessing AI model alignment and scheming
- Evaluate and red-team checkpoints at various stages of post-training
- Automated analysis of post-training data
Role & Responsibilities
The Research Scientist/Engineer will be responsible for running and owning pre-deployment engagements, developing methodology for training-run assessments, and reporting findings to frontier AI labs.
- Run pre-deployment evaluation campaigns with frontier AI labs
- Develop and improve methodology for training-run assessments
- Implement new evaluations and build infrastructure for automated red-teaming
Requirements
- Software engineering skills with Python
- Data analysis and pattern recognition experience
- Excellent writing and communication skills
- Experience using AI to accelerate work
- Market competitive salary
- Equity
#J-18808-Ljbffr…