Product Software Engineer

Build and ship the Windmill product end-to-end. As a developer platform, Windmill's closest comparison is an open-source, self-hostable Cloudflare. The challenge: make it a delightful, intuitive experience while keeping the depth and power developers expect, and make it the best place to author, run, observe, and debug AI-powered workflows and agents.

How we build

Most of our code is now written with AI agents, so writing the implementation is no longer the hard part. The hard parts are deciding what to build and how it should feel, and reviewing what the AI produces deeply enough to catch what is subtly wrong. Building a delightful devtool is the primary goal, and it is hard.

The team is lean by design: a few people with excellent taste ship faster than a larger team. We use herdr extensively to multiplex agent sessions across worktrees, hooked up to Linear and to automatic triage of customer requests, and we expect every engineer to run anywhere from 2 big sessions to 6 smaller ones in parallel. We keep looking for more ways to parallelize work, be more productive, and stay up to date with the latest practices, and we expect team members to push new methods and adapt quickly.

What you'll do

• The editors. Build and refine the script, flow, app, raw app, and pipeline editors in Svelte / TypeScript. These are the heart of the product.
• AI inside the product. Agent steps in flows, AI sessions that build scripts, flows and apps, skills, evals, trace inspection, and the agentic loop behind them. The agent loop and the interface around it are the same product.
• Full ownership. Spec, design, direct the implementation, review it deeply, ship, iterate. Talk to users, ship what matters, cut what doesn't. No handoffs.

Example projects

• Make agent traces actionable: branch or re-run from any step in a past trace
• Extend evals from agents to skills, with eval-gated promotion to production
• Bring the AI agent to the local dev surface: the CLI and the VS Code extension
• Real-time collaboration in the editors, for multiple users and an agent

Who we're looking for

• Excellent product taste in devtools. This matters most. You can tell why one developer tool feels great and another feels like work, you notice the friction others walk past, and you have the judgment to decide what to ship and what to cut.
• Deep review of AI-written code. You read an AI implementation critically, understand it fully, and push back until it is right. That requires knowledge of distributed systems, Rust, PostgreSQL, and Svelte / frontend: at least the minimum in ALL of them, since a single change often touches every layer.
• L-shaped skills preferred. That baseline across the board, plus real depth in one area: devtool DX, AI product engineering, workflow and data infrastructure, or systems.
• AI / LLM skills are highly appreciated. AI is a core part of the product itself, not just of how we build it: agent loops, tool use, evals, prompt and context design.

Portfolio matters more than résumé.

Offer details

• Location: Paris/Geneva hybrid (~3 days/week) or remote within France/Switzerland
• Contract: CDI (permanent)
• Salary: €50K to €95K / CHF70k - CHF105K gross plus 20% bonus on collective milestones. Starting offer based on what you've shipped, top of market for level once you ramp, with quick progression if you prove yourself.
• Also open to: interns / young graduates (5–6 month internship, €2,000–3,000/month, strong CDI potential)