Key Skills

Job Description

Most AI internships ask you to write prompts or do bite-sized changes here and there. This one asks you to rebuild how a deep-tech company works, including proving it worked. Full-time · on-site in Delft · 4–6 months, starting January 2027 · €750/month + commute + vacation days Why this internship You’ll be coached daily by an experienced generalist whose career spans aerospace deep-tech, AI transformation for enterprise clients, and leading the computational modelling at a major nonprofit from inception through >€200M funding to date. You’ll work inside a fast-moving deep-tech startup with more ownership than most interns ever get, on real company infrastructure, with real engineers who will tell you very quickly if something doesn’t work. The learning curve will be steep. If you want six months that move your skill set a few levels up, keep on reading. About us At Emergent we build counter-drone solutions and autonomy systems for teams of drones working together in the real world - defense and emergency response. Bad thinking gets exposed fast in our line of work so we embrace complexity and handle it pragmatically. Small team, high standards, real room to try things and break them intelligently. The role: two streams Stream A - AI Enablement (~60–70%) We’re making Emergent AI-native: redesigning how we work around what AI now makes possible, not just adding AI to existing processes. The groundwork and underlying analysis are done, the first systems are running, and the next phase is yours. Four scoped example sub-tracks (exact tasks will be determined based on immediate priority at the moment of onboarding): A1 - The AI coding harness. Build the scaffolding that makes AI-assisted development reliable: layered instruction files (AGENTS.md / CLAUDE.md as a navigable table of contents, not a monolith), well-designed tools for just-in-time codebase retrieval, mechanical CI invariants that enforce architecture boundaries, a plan-execute-verify loop, and explicit governance - permission separation, human approval gates for destructive actions, cost ceilings. Done right, this is a direct path to ~2× productivity per engineer. Phase 1 is built; phases 2–7 are approved and need to be implemented A2 - Integration testing. Today, getting a build from sim to field takes a myriad of fragile manual steps. Miss one and the field testing day is lost. The fix is disciplined systems engineering - containerization, vertical integration, versioned scenarios, CI-gated integration tests - with AI used to accelerate building the pipeline and to triage failures, not to execute test logic. Target: collapse the number of steps to 1–3. A3 - The AI reviewer. As AI speeds up code production, verification becomes the binding constraint. Phase 2 (yours): take it to production: extend it across the codebase, add cross-repo impact checks, tune it against real reviewer feedback, and measure whether it actually shortens review. Target: reviewers start from checked findings, not a blank page. A4 - The company brain. Critical knowledge lives in people’s heads, chat threads and meetings, and leaves when they do. The fix is a company knowledge base with low-friction capture (automated pipelines from the tools we already use, structured records instead of free-text notes, standard handover files), built to be queried by engineers and AI agents alike. Least friction wins: if it needs a form, nobody fills it in. Phase 2 (yours): connect the sources into one knowledge base that people and agents actually use. All four are the same problem: scattered components and scattered knowledge that need a reliable harness around them. You’ll be the one building that harness. Stream B - Product & Strategy Support (~30–40%) The work a small deep-tech startup always needs: research feeding strategy and product decisions, roadmap input, market analysis, pitch materials, turning messy technical ideas into clear visuals and narratives, supporting demos and external conversations. One week you’re inside a codebase, the next you’re shaping how we pitch it. You’re a fit if you… Black belt common sense and critical thinking: you can tell what matters from what doesn’t Think from first principles and apply Pareto ruthlessly (80% of the outcome, then stop) Understand the basics of coding and how LLMs work well enough to design scaffolding for them: you don’t write production C++, but you know what’s possible and what isn’t Think about verification, governance and failure modes, not just speed Can get busy engineers to adopt something new, sceptics included (especially the sceptics) Handle ambiguity without freezing, and bring structure to messy things Write and communicate clearly (and timely!) Are curious, proactive and want high expectations with real support Are working toward a BSc/MSc in engineering, AI, CS, business informatics, or similar (you must be enrolled in a Dutch university for the whole duration of the internship) Nice to have Hands-on with LLMs, agents, prompt engineering, RAG, or AI-assisted development: built things, not just used ChatGPT Comfort reading code and reasoning about software architecture C++ or Python experience Experience with workflow automation, agent frameworks, or AI tooling (Claude Code, Cursor/Copilot, MCP, LangGraph, n8n) Experience mapping processes (BPMN, architecture diagrams), building demos/prototypes, or strong visual/communication skills Startup, student-team, or project-based experience Practical Strong English required. EU and/or NATO citizenship required due to the nature of our work. How to apply Send your CV and a short motivation note telling us about something you actually made, improved or figured out. If you’ve built anything with LLMs, agents or AI workflows, lead with that.

Core Responsibilities

Build and improve AI-enabled systems and infrastructure, including coding support, integration testing, code review, and a searchable company knowledge base, with attention to verification, governance, and adoption. Support product and strategy work through research, market analysis, roadmap input, pitch materials, demos, and clear communication of technical ideas.

Requirements

Applicants must be enrolled at a Dutch university for the full internship and be pursuing a BSc or MSc in engineering, AI, computer science, business informatics, or a related field; strong English is required. They should have sound critical thinking, basic coding and LLM knowledge, and the ability to handle ambiguity, communicate clearly, and encourage adoption; EU and/or NATO citizenship is required.

Benefits

• Commuting Allowance
• Vacation Days