AROUNDTOWN Commercial Properties (ATCP) is the asset management platform for Aroundtown SA, the largest listed commercial real estate company in Germany. ATCP acts as owner representative thus managing, leasing and optimising the commercial real estate portfolio, which, in addition to typical office properties, also includes industrial, logistics and retail properties. In addition to our company headquarters in Berlin, Aroundtown Commercial Properties holds branch offices in Munich, Düsseldorf, Frankfurt, Hanover and Leipzig. Our employees of various nationalities and their performance stand for the success of our company. Why This Role Exists We are building an AI-native delivery team that ships working software in days, not quarters. This role is for someone who lives in AI-native tooling — Claude Code, Cursor, Replit, v0, Lovable, Bolt — and who treats shipping as the unit of progress. This is not a Staff Engineer role and is not asking for one. We are not asking for five-year architectural ownership; the integration engineering function owns that. We are asking for speed, judgment, and the ability to ship working software end-to-end at high cadence. What Success Looks Like Working software in users' hands fast — days and weeks, not quarters. A portfolio of small, useful internal tools, automations, and tenant-facing PoCs accumulates quickly. Some land, some get killed. Both are signal. Tight feedback loops with real users — internal teams, operations, pilot tenants. Prototypes that prove themselves are handed cleanly to the integration engineering function for hardening — without rewrite drama. Code that did not earn its place is thrown away without ceremony. Responsibilities Building & Shipping Build working software end-to-end in AI-native tooling: Claude Code, Cursor, Replit, v0, Lovable, or equivalent. You are fluent across at least several of these and you experiment with new ones. Ship to small user groups fast — internal users, operations teams, pilot tenants. Your loop is build → demo → learn → rebuild. Cover the stack pragmatically: frontend, backend, basic data, basic deploy. You do not need to be the deepest in any one layer; you need to be effective across all of them. Build LLM-powered features into products where they create real user or operational value — AI-assisted search, document processing, intelligent summarization, agentic workflows. Quality Calibration Write deliberately loose code in early stages; tighten what proves itself. Recognize when something is a five-day experiment and when it is a candidate for production. Throw away what did not work without attachment. Double down on what you did. Hand winning prototypes to the integration & hardening engineering function with clean context — what it does, who uses it, what it touches, what is not yet production-ready. Apply judgment on quality calibration. Prototypes deserve different rigor from production code; both deserve thought. AI-Forward Workflow Use AI coding assistants as your default workflow — code generation, test generation, refactoring, documentation, debugging — not as a novelty. Continuously evaluate new AI development tools as they emerge. Adopt fast when they earn it. Discard when they do not. Maintain critical judgment over every AI output. Speed without judgment produces bad code at scale. Your Profile Experience & Expertise 4+ years of professional development experience or an equivalent demonstrated portfolio of shipped work. Hard requirement: a working portfolio of AI-built prototypes — at minimum 2 to 3 things shipped recently using AI-native tooling. A link is required in the application. Effective across the stack: frontend (React, Vue, or similar); backend (Node, Python, or similar); basic database work; and deployment to platforms like Vercel, Railway, Supabase, or equivalents. Daily practice with AI coding assistants — Claude Code, Cursor, or comparable. Not awareness, not occasional use — default workflow. Working understanding of when something needs to be hardened versus when it should stay in prototype form. Bonus if you have made that judgment call in production before. Comfort with throwaway code and rapid iteration. You do not require comprehensive test coverage on a 3-day experiment. Real estate, proptech, or property operations context is an advantage — not a gating requirement, but it accelerates you. Core Competencies Velocity Under Uncertainty : You ship in days even when the spec is fuzzy. You treat ambiguity as a signal to build, not to delay. Prototype-to-Production Fluency : You know which corners can be cut and which cannot. You know when to call in the integration engineer. AI Tool Mastery : You are deep in at least one AI coding environment and competent in several. You are not a tourist. User Proximity : You stay close to the people using your work. The feedback loop is short and direct. Comfort with Throwing Code Away : You do not get attached. The portfolio matters more than any single artifact. Pragmatic Quality : You apply the right level of rigor for the stage. Experiments and production code get different treatment by design. Who You Are Builder : You measure yourself in things shipped, not story points moved. Allergic to Long Planning Cycles : You would rather build a small bad thing on Tuesday than plan the perfect thing for two weeks. Comfortable with Messy Code in Service of Learning : You also know when to clean it up — or throw it out entirely. Curious About the Domain : You actively want to understand how a property gets managed, what an operations team does day-to-day, and where the friction lives. AI-Forward : AI tools are your default, not a novelty. You experiment with new ones every month and have opinions about them. Discreet : You handle tenant data and operational information responsibly, including data passing through AI systems. Qualifications Working portfolio of AI-built artifacts shipped in the last 90 days — link required in a