About Scrimdata
Scrimdata builds frontier RL environments from real company data. We license the operational history of real companies, run it through our own anonymization pipeline, and turn it into multi-step, multi-tool training environments: tasks, rubrics, verifiers, and reference trajectories, graded against what actually happened. We're an early-stage, remote-first team selling to two sides of one market: AI labs and enterprise agent teams who train on our environments, and data partners who supply the raw material. Small team, hard problems, real customers from day one.
About the role
We're hiring a Staff Product Engineer to turn a fast-moving, mostly manual pipeline into product our customers can actually use themselves. Right now, browsing a catalog, inspecting a task's rubric, replaying a trajectory, or checking why a batch of rewards looks off means someone on our team digging through logs and spreadsheets. That's the job: build the tools that make our environments legible, browsable, and trustworthy to the researchers and post-training teams who use them every day. You'll work close to actual customers, watch how they actually work, and decide what to build next. Nobody is handing you a spec. You're finding the problem and shipping the fix.
What you'll do
- Design and build the interfaces our customers and internal team use to browse, filter, and select environments and tasksets
- Build task and rubric inspection tools: let a researcher see exactly what a task asks, how it's graded, and why a reward landed where it did
- Own trajectory review UX end to end: replaying an agent's run against a reference trajectory, surfacing where and why it diverges
- Build quality dashboards that make environment health visible: coverage, difficulty distribution, verifier reliability, reward signal noise
- Talk directly to post-training and evals teams at partner labs, understand what's slowing their pilots down, and turn that into a shipped tool, not a ticket
- Prototype fast, throw away what doesn't work, and take the parts that do to production quality without a long design queue in between
- Work across the stack as needed: frontend for the tools themselves, backend and data pipeline where the product problem lives there instead
What we're looking for
- 7+ years of production software experience, with real ownership of a product from idea through to something people use daily
- A product-first mindset: you'd rather sit with a user for an hour than guess at what they need, and you can turn that hour into a working prototype fast
- Fluency across a modern stack: comfortable in Python and TypeScript/React, and not precious about which layer the fix belongs in
- Track record of shipping in ambiguity: you've been the first engineer on a feature, a team, or a company, and you know how to make the call without waiting for someone else to make it
- Strong communication. You'll be in the room with technical customers, translating what they say they want into what actually solves their problem
- Comfort with a small, fast-moving team where the roadmap changes as fast as what we learn from the last customer conversation
- A CS degree from a strong program or an equivalent, demonstrated STEM background
Nice to have
- Early-stage startup experience, as an early engineer or a founder
- Direct exposure to RL post-training, evals, or agent-training workflows, from either the lab side or the tooling side
- Experience building internal tools or data products where trust and legibility mattered as much as raw functionality
- Comfort working with large-scale, messy real-world data and the pipelines that clean it up
Benefits
Meaningful early equity. Remote-first, work from wherever you do your best work. Standard benefits package.
How to apply
Email careers@scrimdata.com with your resume and something you've built that people actually use. No forms, no links.