About the job
Application Deadline: We are actively conducting interviews and aim to onboard the right candidate promptly.
ABOUT THE ROLE
At Apollo Research, we are embarking on an ambitious journey to establish a groundbreaking field known as the “Science of Scheming.” We are seeking passionate Research Scientists and Engineers eager to contribute to the creation of a new hard science from the ground up.
YOUR RESPONSIBILITIES
- Collaborate with leading AI innovators. Our partnerships with various labs provide access to an extensive range of models that no single AI lab can offer. Your contributions will significantly influence the development and deployment of advanced AI systems.
- Conduct in-depth studies of reinforcement learning dynamics that drive the emergence of reward-seeking behaviors, evaluation awareness, and misaligned preferences. Design and train model organisms, scaling insights to cutting-edge systems.
- Explore the “Scaling Laws of Scheming.” Establish empirical foundations to forecast how scheming risks evolve as models increase in capability.
- Innovate evaluation techniques that can potentially scale to highly evaluation-aware models.
- Investigate AI cognition. Uncover unique patterns in the reasoning processes of advanced AI systems that have yet to be documented.
Note: This position does not focus on interpretability roles.
KEY REQUIREMENTS
We seek a diverse set of skills to advance our research agenda. We do not expect any single candidate to meet all the criteria listed below. However, a successful candidate will likely excel in one or several of the following areas:
- Fast-paced empirical research: Ability to design and execute experiments with a focus on accelerating iteration cycles.
