Procurement is the single largest cost lever in manufacturing: over 50% of a physical product's cost originates with suppliers. Yet the teams managing billions in spend still run on spreadsheets, fragmented tools, and tribal knowledge.
Delvo is building Cognitive Procurement: an AI-native intelligence layer that gives strategic procurement teams the decision power they've never had. Our agents analyze suppliers, benchmark prices against real-time cost indices, and generate tailored negotiation strategies, turning weeks of manual analysis into minutes of actionable insight.
We're founded by a McKinsey venture builder who shipped production AI systems as interim CTO, a procurement excellence lead from PwC who ran 500+ supplier negotiations for industrial manufacturers, and one of the very few people in the world who combines deep procurement domain knowledge with the technical skills to build. We combine engineering depth with domain expertise to build something procurement teams actually trust and use.
If you want to define how humans and AI work together in enterprise procurement, this is the role.
We're looking for a Design Engineer: someone who designs interfaces, talks to users, and prototypes in code so ideas can be tested before they're built. Not a designer who hands off Figma files. Not a frontend engineer who needs a spec. Someone who thinks in user flows and validates them with working prototypes, and who has strong opinions about how AI products should feel.
The most important question at Delvo right now isn't "what color should this button be." It's: what does the right interface for AI-powered procurement even look like? Maybe it's not a dashboard. Maybe it's a conversational interface in Slack with interactive charts, inline actions, and real-time data cards. Maybe it's a voice briefing that prepares you before a supplier negotiation. Maybe it's something nobody has built yet. We need someone who thinks about these questions obsessively, prototypes answers in code, and validates them with real users.
You'll work directly with our CPO and sit in on customer calls every week. You'll design, prototype in code, test with users, iterate, and ship. The loop is tight and the impact is immediate.
Define the interaction model for AI in procurement. This isn't about skinning an existing product. The interface between an AI agent and a procurement professional barely exists yet. You'll invent it. The patterns you establish will define how an entire category of software works.
Design with real users, not personas. You'll talk to category managers, procurement directors, and CPOs at major industrial companies every week. You'll watch them use the product. You'll understand their workflows, their frustrations, and the moments where AI can transform a decision. Your designs will be grounded in reality, not assumption.
Make LLMs feel magical. Our agents do complex things: analyze thousands of supplier data points, benchmark prices across global indices, generate negotiation playbooks. Your job is to make all of that complexity disappear. That means designing interfaces where AI output isn't just text in a chat bubble, but interactive charts you can drill into, action buttons that trigger real workflows, data cards that update in real time. The user should feel like they have the world's best procurement analyst sitting next to them, not like they're operating a complex system.
Prototype at the speed of thought. You'll go from sketch to functional code prototype in hours, not weeks. When you have a hypothesis about a new interaction pattern, an interactive chart component, or an agent-driven workflow, you build it, put it in front of a user, and learn. No handoff delays, no waiting for engineering capacity. You are the capacity.
From prototype to product. Your best prototypes will inform what gets built into production. You'll work closely with our engineers to translate your validated designs into the real product. If you can write production-quality frontend code yourself, even better, but the core expectation is that your prototypes are functional enough to test real interactions, not static mockups.
You should:
Be a strong designer who prototypes in code. You think in user flows and interaction patterns, and you can build functional prototypes in React/Next.js to test them with real users. You're not afraid of a codebase. If you can also write production-quality frontend code, that's a significant bonus, but the core requirement is that code is your prototyping medium, not just Figma.
Have a portfolio that shows original thinking about interaction design, not just polished visual execution. We want to see how you think about problems, not just how you style solutions.
Be obsessive about user research. You know how to run a discovery call, pull insights from messy qualitative data, and translate what users say (and what they don't say) into product direction. You'd rather spend an hour watching a user struggle than a day debating layouts in Figma.
Have a point of view on how AI products should work. You've thought about conversational interfaces, agent-driven UX, proactive AI, or ambient intelligence. You have opinions about what makes AI features feel trustworthy versus gimmicky, and you can articulate why.
Be comfortable challenging assumptions. If the obvious answer is "build a dashboard," you should be the person who asks "but should we?" The best interface might be a Slack integration, a voice briefing, or something entirely new. We want someone who explores the full design space before converging.
Move fast and iterate. You prototype in code, not in committee. You're comfortable showing rough work to users early and learning from their reactions instead of polishing in isolation.
Bonus: experience designing data-dense or analytical interfaces. Our users work with cost data, supplier portfolios, and market indices. Making complex information clear and actionable is a real skill.
Bonus: any exposure to B2B, enterprise, or workflow-heavy products.
Design: Figma for exploration and stakeholder communication. But your primary design tool is code.
Prototyping: React/Next.js, Tailwind. You build functional prototypes that users can actually interact with: click through flows, see real data visualizations, trigger actions. Static mockups don't test what matters. v0, Cursor, Claude Code, Codex, whatever it takes.
Research: Whatever works: live user sessions, Loom recordings, usage analytics, direct customer calls. The method matters less than the insight.
We're a small, fast team with large-enterprise customers and the intensity that comes with it. We work from CIC Berlin in Kreuzberg: direct collaboration, no unnecessary process.
Everyone on the team talks to customers. Everyone ships to production. The founders are technical and hands-on: Jonas (CTO) writes code daily, Darius (CPO) designs and builds product, Erik (CEO) runs customer relationships. There's no layer between you and the work that matters.
We're building a generational company in Cognitive Procurement. That means high standards, fast pace, and real ownership. You'll define how procurement teams experience AI, not just how it looks, but what it feels like to make better decisions.
💰 Competitive salary with meaningful equity (VSOP). You're joining at the earliest stage and your compensation reflects that.
🏖️ 30 days vacation.
🖥️ Equipment budget for your ideal setup.
Search for a command to run...