Applied AI Engineer

About Us

2,100+ customers across 80+ countries trust us, including Cleary Gottlieb, Goodwin, Linklaters, White & Case, Dentons, and Barclays. We’ve scaled to $200M+ in ARR, with teams across Europe, North America and APAC, and continue to expand through acquisitions including Qura, Walter AI, Graceview, Cadastral, and Wexler.

Stockholm | Full-time | Onsite, five days a week

The team

AI Enablement makes Legora AI-native by default. We work across three streams in parallel: building internal agents and tools, equipping teams to build their own, and owning the standards, access and guardrails that let everyone move fast safely.

We own the AI platforms and the internal products built on them. So far that includes Leya, an internal knowledge agent live in Slack that answers questions about Legora with cited sources and automates parts of our RFP and security questionnaire work. A structured, agent-readable knowledge base assembled from knowledge each domain continues to own, where agents propose changes and humans approve them.

Leyf, a platform for internally built apps that lands anyone's app on shared infrastructure with guardrails scaled to how much damage it could do. And a Skills Hub, so that something one person works out becomes something the whole company can use rather than staying on their laptop.

The team is small and the mandate is large. That is the job.

What makes this hard

Legora is an AI company of around a thousand people, most already fluent and on current tools. Internal AI here is not an adoption problem. Nobody needs convincing, which means there is nowhere to hide behind change management or training. If something we build is not used, it is because it was not good enough.

Our colleagues know what good looks like and will not use something worse than what they think they could build themselves in an afternoon. So we ship things people genuinely want, and the foundations arrive with them. That constraint does not go away.

The mandate is also larger than the team. The part no individual can do for themselves is the infrastructure, the access model and the architecture underneath, and getting that right across a company this size is the actual work. It does not come with a specification.

What you will do

You will take broad ownership of how AI works inside Legora and of the platforms we build to make it work.

• Build internal agents, tools and platforms, and treat them as products rather than projects. Quality, reliability and whether they actually serve the business are yours, not someone else's
• Take a broad goal, work out what is actually required, and break it into the work yourself
• Own what you build end to end, from the architecture through to whether people use it and it holds up in production
• Shape what we should build and in what order, and argue for it
• Design the foundations: knowledge architecture, access and permissions, evaluation and monitoring, the guardrails that scale with risk
• Work directly with the people who will use what you build, across the whole company, including the parts furthest from engineering

Who we are looking for

• Strong engineering fundamentals. You build production systems, not prototypes that need someone else to finish them
• Real depth with LLM-based systems. Agents, retrieval, evaluation, tool use. You know where these systems break and you design for it
• Ownership by instinct. You do not wait for a specification. When the requirement is unclear or wrong, you say so and propose something better
• Three kinds of understanding at once. Technical, domain and business. You can hold a conversation about architecture and a conversation about what the business actually needs, and connect them
• Openness. You take feedback, build on other people's ideas, change your mind when the evidence moves, and push for the strongest outcome rather than the one you proposed first
• Comfort with ambiguity. New team, fast growth, decisions that get revisited as we learn. That should sound appealing rather than annoying
• TypeScript. It's what we work in, and where our shared patterns live
• No aversion to infrastructure. We own ours, so Terraform and cloud providers are part of the work rather than someone else's problem
• You work with code agents daily. Not as an experiment, as how you build