Open position
Member of Technical Staff (AI Researcher)
Study how frontier models develop cyber capabilities and how to measure and improve them.
The role
Design experiments that explain what frontier models can do in realistic security environments. You will work on agent evaluations, reinforcement learning, and the relationship between training performance and held-out capability, alongside researchers who understand the underlying security tasks.
What you’ll work on
- Develop evaluation methods for offensive and defensive agent capabilities, including trial design and uncertainty estimates.
- Investigate agent behavior, failure modes, generalization, and sensitivity to tools and inference budgets.
- Build reinforcement learning experiments using realistic environments and independently verified rewards.
- Study reward hacking, benchmark contamination, and the separation of training tasks from held-out evaluation.
- Write reproducible experiment code and technical reports that explain both findings and limitations.
What you’ll bring
- Strong foundations in machine learning, experimental design, and statistical reasoning.
- Experience conducting empirical research with language models, agents, reinforcement learning, or related systems.
- The ability to implement and debug experiments in Python and modern ML frameworks.
- Evidence of independent research judgment through projects, publications, or other substantial technical work.
Compensation & benefits
This is a remote role with equity, healthcare, and other benefits. We’ll discuss the compensation package and benefit details with you during the hiring process.
How to apply
Email a short introduction and your résumé or profile to careers@counterweightlabs.com. Put Member of Technical Staff (AI Researcher) in the subject line. A project, research write-up, or other example of your work is welcome; please only share material you’re authorized to share.