Senior LLM Research Scientist

Senior LLM Research Scientist

A frontier-stage research group is building a new class of AI systems designed to reason, plan, and act across the physical world. Their mission is to create intelligent agents capable of experimenting, engineering, and constructing in ways that dramatically accelerate scientific and industrial progress. This team combines deep technical pedigree with real-world wins at scale, including major government-funded initiatives. They operate where advanced model research meets robotics, simulation, and automated engineering systems, offering the kind of impact only possible when first-principles science meets ambitious execution.

Joining means stepping into a high-ownership environment where you shape core capabilities end-to-end, influence the direction of physical-world intelligence, and help build technology the world has never seen before.

Why This Role Is Compelling

• Work on cutting-edge reasoning, planning, and tool-use models that directly control autonomous engineering systems.

• Push the limits of SFT, RLHF, DPO, verifier-guided RL, and long-horizon planning in a setting where your research immediately translates into real-world capability.

• Operate in a high-velocity research culture with exceptional peers across agent systems, simulation, data, and complex tool chains.

• Have outsized ownership in a small team tackling one of the most ambitious technical problems of this decade.

Role Overview

The team is looking for an LLM Research Scientist to pioneer next-generation reasoning and agent architectures. Your work will span model design, alignment strategies, structured tool orchestration, and experimentation with agents interacting across real engineering workflows. This position blends deep research with hands-on systems integration, offering both autonomy and scope to lead foundational progress.

Key Responsibilities

• Develop advanced models and prompting systems for planning, multi-step reasoning, and structured tool use.

• Lead training initiatives across SFT, RLHF/DPO, verifier-guided RL, and modular expert architectures to strengthen robustness and cont rollability.

• Define schemas, tool-calling strategies, policy constraints, safety mechanisms, and recovery pathways for agent behavior.

• Partner closely with engineering, simulation, and data teams to test, train, and evaluate models embedded in real production-like tool chains.

Qualifications

• Significant experience in LLM research, agent reasoning models, or structured tool-use frameworks.

• Strong background working with SFT, RLHF, DPO, or reinforcement-learning-from-verification methods.

• Demonstrated ability to design, analyze, and improve long-horizon behaviors and decomposition strategies.

• Comfortable working across ML research, systems engineering, and real-world experimentation in a fast-moving environment.

• A track record of excellence and ownership in technically demanding domains.

#JLjbffr