Position
Research Scientist, Discovery (Engineering & Applied Sciences)
Anthropic
- San Francisco, United States · Hybrid
- Artificial Intelligence, Reinforcement Learning and
About
Anthropic’s Discovery team is organised around building an AI scientist: a system that can carry the long-horizon reasoning needed to push the scientific and technological frontier. This role extends that work to engineering and the applied sciences, training Claude to invent, design and engineer new manufacturing and experimental techniques, processes, materials, devices, instruments, machines and energy systems. The research scientist designs reinforcement-learning environments, rewards, training data and benchmarks for realistic engineering and scientific workflows; builds simulation-based environments and verifiers with domain experts and validates them against real-world data; studies long-horizon task completion, multi-agent collaboration, complex reasoning and learning from experimental feedback; and scales research ideas from prototype to production training runs.
The annual salary range is $500,000–$850,000 USD.
Requirements
Proficiency in Python and hands-on experience with a modern machine learning framework such as PyTorch or JAX; hands-on experience training or evaluating machine learning models; a track record of machine learning research (forming hypotheses, designing and running experiments and drawing sound conclusions); ability to implement research ideas and diagnose problems in complex systems; ability to learn unfamiliar technical fields quickly and work with experts in them; clear communication. A background in engineering or the sciences is a plus but not required. Anthropic expects staff to be in one of its offices at least 25% of the time. Minimum education: a bachelor’s degree or an equivalent combination of education, training and experience.
Topics
- AI scientist, search World Wide
- reinforcement-learning environments, search World Wide
- Claude, search World Wide
- reinforcement-learning, search World Wide
- long-horizon reasoning, search World Wide
- multi-agent collaboration, search World Wide
- simulation-based environments, search World Wide
- experimental feedback, search World Wide
- benchmarks, search World Wide
Source: Anthropic