about 4 hours ago
Responsibilities
- Own one hard AI problem from proposal to production.
- Build systems and design evaluations based on real production data.
- Engage in projects related to harness and inference-time work.
- Develop post-training strategies for agents and design reward systems.
- Create training and evaluation environments from enterprise workflows.
- Mine production data for training corpora and evaluate agent performance.
Requirements
- No specific degree required; demonstrated depth in ML or agent systems is essential.
- Strong fundamentals in machine learning and engineering skills in Python and PyTorch.
- Real depth in at least one area such as post-training, reward modeling, or agent systems.
- Statistical literacy and the ability to size experiments accurately.
- A habit of honest measurement and a preference for clean experiments.
Benefits
- Flexible start dates and half/full-time options.
- On-site/hybrid work arrangements in the SF Bay Area.
- Competitive salary of $4,000 per month.