Grafton Sciences is looking to solve the problem of developing physical general intelligence through autonomous systems that can experiment, reason, and discover in the physical world. The Senior LLM Research Scientist will contribute to this by developing advanced planning, reasoning, and tool-use models for autonomous engineering systems.
Requirements
- Deep experience in LLM research, reasoning models, agent systems, or structured tool-use frameworks.
- Strong background in SFT, RLHF, DPO, or verifier-driven RL, with a track record of pushing model capabilities.
- Ability to design and evaluate long-horizon behaviors, decomposition strategies, and structured policy constraints.
Responsibilities
- Develop models and prompting frameworks for planning, tool use, and long-horizon reasoning.
- Lead SFT, RLHF/DPO, verifier-driven RL, and MoE routing strategies for robust agent behavior.
- Design schemas, tool-call policies, safety constraints, and failure-recovery mechanisms for LLM agents.
- Collaborate with agent systems, data infrastructure, simulation, and tooling teams to train and evaluate models in real workflows.
Other
- Comfortable working across ML, systems engineering, and real-world toolchains in a fast-moving research environment.
- Above all, we look for candidates who can demonstrate world-class excellence.
