[Hiring] Builder @Vinmar International

Role Description

A Product Led Builder owns a product area end to end — discovery, decisions, and delivery — and ships the product themselves using AI agents. This is not a PM who writes specs and hands them off. It's a PM who builds.

• The spec is the prototype. The prototype becomes the PR.
• When AI can generate working code from a precise brief, the slowest step in product development is the handoff between the person who understands the problem and the person who writes the code. This role removes that handoff.
• You'll sit with sales reps, pricing managers, and product directors across our tenants, figure out what actually slows them down, and put working software in front of them — fast.
• Engineering reviews and hardens; you own the outcome through to production.
• Our users are busy people doing demanding work. A rep closing 5–15 deals a week needs tools that keep pace with them.
• A pricing manager who spends a third of their week tracking feedstock and market news benefits most from clear, timely answers.
• You'll build for them directly, on surfaces like Catalog, PriceBook, order approvals, and the opportunity tracker.

We ship every week.

• Builders at Vailent put something into production every week. Not a demo, not a deck — a change real users can touch: a fix, a slice of a larger feature, or an instrumented experiment.
• Our users move at the pace of 5–15 deals a week, and the product should keep up with them.
• Slice big bets thin. Larger features get broken into increments that each reach production on their own, behind a flag if needed.
• Every ship is demonstrated. A working screen, a read-back record, or a live payload in the real environment — then a handoff note so the next person starts informed.
• Missed weeks get named. If a week ends without a ship, you say so plainly, say why, and say what unblocks the next one.

What You'll Do

• Own a product area's outcomes. Discovery, prioritization, and delivery for a surface like PriceBook or Catalog. Three months in, we should be able to point to things that exist because you built them.
• Ship to production every week. Keep a steady weekly cadence of production changes in your area, each one demonstrated with users or in the live environment — momentum is part of the job, not a bonus.
• Go to the user first. Run discovery directly with sales reps, customer service reps, pricing managers, and product directors across our current and future customers. Map the real workflow — the spreadsheet, the SAP screen, the email chain — before proposing anything.
• Prototype before you debate. Turn a problem into a working prototype in days, not a PRD in weeks. Use it to align stakeholders and test with real users.
• Ship production code with AI agents. Use Claude Code and agentic tooling to generate, test, and open production-ready PRs against the actual repo. Engineering reviews and hardens; you carry it to done.
• Write specs precise enough for an agent. When you do hand off, your brief is tight enough that an AI agent or an engineer can act on it without follow-up questions.
• Respect the data model. Read our schemas and ERP-mapped master data well enough to judge whether AI-generated scaffolding is correct — customer IDs, sales areas, material groups, tenant rules.
• Close the field-to-product loop. Turn support tickets, CSAT signals, and integration issues surfaced by Customer Success into prioritized, shipped fixes.
• Measure what you ship. Define success before you build, instrument it, and report back plainly on what worked, what didn't, and what you cut.
• Product judgment first. You've owned product outcomes — not just a backlog. You can explain why you built something, what you chose not to build, and how you knew it worked.
• You've already crossed the line into building. You've used Claude Code, Cursor, or similar tools to ship real software that real people use. Not a weekend demo — something in production. Show us.
• A real technical floor. You don't need a CS degree, but you can read a data model, follow an API contract, run the app locally, and tell when AI-generated code is wrong.
• The evidence reflex. "Done" means demonstrated. You prove things with a working screen, a read-back record, or a live payload — never a green badge.
• Comfort with messy, enterprise data. You've worked with systems that weren't designed to talk to each other — ERPs, CRMs, spreadsheets, flat files — and you know that's where the product problems live.
• Customer credibility. You can sit across from a commercial director at a chemicals distributor, earn their trust, and speak their language plainly.
• High agency, low ego. You push through blockers instead of escalating them, take blunt feedback well, and reverse your own conclusions out loud when the evidence turns.