What You'll Do- Pair with an FDPM on every project for a given product category, from the first build through every deployment after it
- On a new product type, design and build the components, data models, and workflows as reusable scaffolding
- On later deployments of a product type, extend and harden that scaffolding against each new customer's specific data and edge cases, while the FDPM handles configuration and finishing touches
- Get your hands dirty with real government data: legacy exports, mainframe extracts, undocumented APIs, and the inevitable exceptions that come with it
- Use AI coding tools to build and iterate quickly, shipping working modules against a live customer instead of a backlog
- Feed what you learn about the platform's gaps back to the Platform Engineering team so the lower-level primitives improve with every product category you build
What We're Looking For- A portfolio: what have you built end-to-end, from scratch, under pressure?
- Strong backend fluency, with good instincts for abstraction: knowing what to generalize into reusable scaffolding vs. what to leave specific to one customer
- Enough frontend ability to build and ship a usable UI without needing to hand it off
- Hands-on experience using AI coding tools (Claude Code, Codex, Gemini CLI, Factory, etc.) as a core part of your daily workflow
- Comfort working directly with non-technical stakeholders alongside an FDPM, even if you're not the one owning the relationship
- High-agency, high-velocity mindset: comfortable being the first engineer into a brand-new product category, and just as comfortable staying paired in on every deployment after it
- US citizen (required for government work)
- Willing to work in-person in Flatiron, NYC (5 days/week)
Nice to have: experience building internal tooling, developer platforms, or reusable component libraries; prior startup experience shipping directly to customers; public sector domain knowledge (courts, permitting, utilities, school districts); or experience working with legacy/mainframe data.