Member of Technical Staff - London - UP TO £250k + Equity

Member of Technical Staff - Pre Seed Start Up Up to £250k + Equity On Site in London Frontier models are great at answering questions. They're much worse at behaving like real people, systems or environments over time. This team is fixing that. I'm working with a London AI start-up building models that power high-fidelity simulations. Instead of relying on off-the-shelf frontier models, they're post-training their own to act, react and reason realistically inside simulated environments, so customers can test decisions before making them in the real world. They're hiring a Member of Technical Staff to own post-training and RL, turning research ideas into models that make the simulations sharper. What you'll do

• Build post-training pipelines (SFT, preference tuning, RL) that make models behave more realistically in simulation
• Design reward signals and environments for RL, and figure out what “realistic” actually means in measurable terms
• Build evals that prove a trained model beats the frontier baseline on the tasks that matter
• Run experiments fast, read the results honestly and ship what works into production
• Help set the research and engineering direction as the team grows What they're looking for
• Hands-on post-training experience on real models (GRPO, PPO, DPO, LoRA or similar)
• Someone who's built RL environments, reward models or evals, not just run training scripts
• Strong engineering fundamentals: you can scale training and keep it reproducible
• Comfort with ambiguity and a habit of turning open research questions into shipped results Why it's worth a look
• RL for simulation is a genuinely open problem, so your work moves the frontier, not just the product
• Real ownership of the model layer at an early stage
• A small, technical team where research goes straight into production