SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Multiverse is Europe's first EdTech double unicorn, recently valued at $2.1bn after raising $70m in strategic funding. The company partners with 1,500+ organizations to deliver upskilling apprenticeships in AI, data, and tech skills, with learners driving $2bn+ ROI for employers.
You will join the AI Transformation team, a small, focused squad tasked with rebuilding how Multiverse operates function-by-function to become an AI-first company. This is not about bolting AI onto existing processes—it's about fundamental transformation that will serve as a blueprint for the broader UK tech sector and economy.
In this role, you will own complete agent systems end-to-end: from product problem definition through architecture, implementation, evaluation, and production operation. You will design context and retrieval strategies, determining what enters the context window and what stays out—the most consequential design decision in AI systems. You'll build evaluation frameworks that measure accuracy, safety, helpfulness, domain-specific quality, and latency, treating evaluation as an engineering discipline.
You will design and build the tool integration layer (MCPs, APIs, data contracts, error handling) that enables agents to reliably interact with Multiverse's wider product ecosystem. You'll influence technical direction through evidence-backed opinions, contribute to architectural decisions, and raise the bar through rigorous code review and pairing with less experienced engineers.
Your primary development workflow will use Claude Code, where you set context, define constraints, critically review output, and augment the tool with domain expertise. You'll work closely with engineering teams in London and Berlin, integrating existing systems while building new ones from scratch.
You have shipped production AI systems serving real users at meaningful scale. You understand context management, model selection and routing, and cost engineering. You've made trade-offs between context quality, latency, and cost in production environments. You're comfortable with multiple models and understand when smaller, faster models are the right choice. You know token economics, caching, prompt optimization, and batching.
