You will own difficult AI and research-engineering problems from initial question through experiment, implementation and production impact. This includes designing approaches for creating, curating and improving high-quality training data, fine-tuning and post-training models using methods such as supervised fine-tuning, reinforcement learning and automated prompt or harness optimisation, and building the evaluation and feedback loops that make model and agent performance measurable and continuously improvable. You will also work on agent harnesses, sandboxed environments and the systems that connect models to interactive software tasks.

Working closely with software engineers and curriculum engineers, you will prototype quickly, identify what actually works and turn successful approaches into robust, observable production systems. At Staff / Principal level, you will define technical roadmaps for major areas, set direction across agents, evaluation, training data and model improvement, and help other engineers structure ambiguous AI problems.

Requirements

• Senior, Staff or Principal-level AI / ML or research engineer with deep hands-on technical ability.
• Direct experience fine-tuning or post-training models, rather than only consuming model APIs.
• Strong understanding of modern AI / ML fundamentals, including model training, optimisation and evaluation.
• Practical knowledge of supervised fine-tuning, reinforcement learning, RLHF or related post-training methods.
• Strong Python and the software-engineering depth to write clean, testable code.
• Proven ownership of meaningful technical outcomes across research, data and engineering trade-offs, solving ambiguous problems from first principles.
• Experience with agents, agent harnesses, sandboxed environments, evaluation infrastructure or synthetic-data workflows is highly valued.

Compensation & Benefits

• Base salary of approximately £100,000–£240,000, depending on level and experience.
• Meaningful equity in a well-funded, rapidly growing frontier AI company.
• Hybrid working in London or Paris, typically two days per week onsite.
• The first five weeks are expected to be full-time onsite in your local office.