Alignment Scientist/Engineer
You'll work autonomously in a fast-paced, high-stakes environment as a high-agency technical generalist. You'll guide the direction of research alongside external collaborators, working on ambitious alignment directions such as pretraining interventions for model safety, novel interpretability techniques, and neuroscience-inspired approaches to alignment. You'll call AI model APIs, build and train deep learning models, and use AI tools to automate your own workflow.
Responsibilities
- Guide the direction of research alongside collaborators
- Research pretraining interventions for model safety
- Develop novel interpretability techniques
- Explore neuroscience-inspired approaches to alignment
- Build, train, and evaluate deep learning models
- Call AI model APIs via REST or client libraries
- Decompose complex problems into concrete, testable experiments
- Communicate technical work to all audiences
Requirements
- Experienced in ML research, alignment research, or both
- Deeply passionate about AI alignment and belief that genuine progress is possible
- Excellent at Python, Pytorch, and the modern deep learning stack
- Knowledgeable about building, training, or evaluating deep learning models
- Comfortable calling AI model APIs via REST or client libraries
- Adept at decomposing complex problems into concrete, testable experiments
- AI-native, and eager to use AI tools to automate workflow
- Thrives in a fast-paced, startup-like environment
- Adept in English and comfortable communicating technical work to all audiences
Benefits
- Remote-first with flexible schedules
- Health insurance
- Free lunches at the office
- Weekly knowledge-sharing talks
- Annual team retreat
- Regular performance-based raises
- Diversified employee equity program