We are Juris Legal. We build an AI product for the legal world, which means we have to be unusually good at applying AI to our own work. This is a build-heavy internship, not a research or observation role. You will ship internal tools, automate real processes, and be accountable for whether what you shipped still works a month later. The bar You should already have built things with AI tools, not just used them. If you cannot point to something you made, a script, an automation, an agent, a pipeline, an internal tool, this is not the right role yet. We care much less about which programme you are on. Concretely, you can: write and debug your own code (Python, JavaScript/TypeScript or similar) work with APIs, and wire tools together into something that runs unattended judge an AI output critically: know when it is wrong, and how you would catch it at scale What you will do Test new AI tools and workflows as they land, and decide fast what is worth our time Apply them to real internal processes, ours, in production, with real constraints Build and maintain small prototypes and utilities: scripts, automations, internal tools Turn what works into something the team can rely on: short docs, playbooks, a demo Evaluate against clear criteria: quality, reliability, cost, privacy, security What success looks like By the end, a few things you built are still in daily use by the team, and we make better adopt, pilot or avoid calls because of work you did. Practical Copenhagen. Flexible hours, remote work possible. Start by agreement, from autumn 2026. Minimum three months. We fit the scope and hours to your programme's requirements. Unpaid. The placement must be a formal part of your education. We review applications on a rolling basis. Applying Answer the application questions and include a link to something you have built. A link beats a CV for this role. We read every answer to the written questions, so keep them concrete and specific to your own work.