Software Product Engineer - LATAM

COOPr · Remote, LATAM-based · Contract · Mid-level (5+ years)

COOPr is an operations platform for the construction industry. We're using invisible data capture mechanisms to bridge the gap between the field and finance and help MEP (Mechanical, Electrical, Plumbing) subcontractors accelerate their cash conversion cycle. We build upon our labor, project progress and material modules to create billing packs, and we have since helped data center subs secure $MM in previously locked payments.

We're looking for a product engineer who will live in the codebase day-to-day: shipping features end-to-end, fixing customer-reported bugs fast, owning the evolution of our data models, and being a sharp counter-balance to AI-generated code. This is a contract role, calibrated to a LATAM time zone for meaningful overlap with our NYC office and US customers.

Ship features end-to-end. Schema change → API → frontend → instrumentation → in front of users. You won't hand off pieces; you'll own the whole arc and make the trade-off decisions along the way.

Own and evolve our data models. You'll own the evolution of a large, real-world schema in a multi-tenant data model. That means owning migration safety, denormalization decisions, index strategy, and the long tail of "this column should have been NOT NULL three months ago" work.

Triage and fix customer-reported bugs fast. You'll hold a morning support block dedicated to bug triage and hotfixes — typically 2-3 hours of your morning, which overlaps with US business hours. After that block, you're in feature flow. We're explicit about this carve-out so feature work doesn't evaporate.

Build AI-native features. AI-powered features are central to how the product creates value — from the in-app assistant to automated document generation to turning unstructured field inputs (voice notes, photos, scanned forms) into structured records the team can act on. You'll ship new AI features end-to-end and harden the surrounding infrastructure that keeps them reliable in production.

Work alongside the CEO and other engineers in an AI-assisted flow. Cofounders ship features daily. You'll be reviewing AI-augmented PRs as part of normal workflow: catching missing instrumentation, tenant-isolation gaps, untested error paths, and schema-migration risk. Pushing back when something looks wrong is part of the job, not an exception.

Refine the product for scale across multiple customers. As we sign more pilots, the difference between "works for our design partner" and "works for any contractor in the field" lives in the details. You'll partner with the cofounders to find and close those gaps.

• Morning block (~3 hrs your local time): bug triage, customer issues, hotfix work
• Afternoon: feature flow on planned roadmap work
• 4-5 hours of overlap with US Eastern time for syncs and pairing
• Daily product sync with the CEO

• Backend: Node.js + Express.js, TypeScript (strict), service-oriented architecture
• Database: Prisma ORM + PostgreSQL on Google Cloud SQL
• Auth: Firebase Authentication with multi-tenant org/project scoping
• Infrastructure: Google Cloud Run, Cloud Build CI/CD, Cloud Storage, Cloud Tasks
• Real-time: WebSockets, Deepgram transcription
• AI: Anthropic Claude SDK, Google Cloud Vision
• Comms: Twilio (SMS/voice), WhatsApp webhooks, Nodemailer, Web Push
• Integrations: Procore, Google APIs, Browserbase

• Have 3+ years shipping production software as a full-stack or backend-leaning engineer.
• Are fluent with AI development tools (Claude Code, Cursor, GitHub Copilot, or equivalent) and can articulate concretely when they get something wrong and how you caught it. Generic "yes I use AI" answers don't count.
• Have hands-on Prisma + PostgreSQL experience, including the unglamorous parts: shadow databases, migration safety on large tables, index strategy, and "I've been bitten by this before" stories.
• Have worked on multi-tenant SaaS where org/tenant scoping is a first-class concern, and your first instinct on any endpoint is "is this properly scoped?"
• Are comfortable owning data models end-to-end — including making backwards-compatible schema changes on hot tables under real traffic.
• Can ship a hotfix in under 24 hours without losing your composure, and also do focused feature work in the same week.
• Write clearly in English — PR descriptions, written reasoning, async disagreements. We'll ask to see a public artifact (PR thread, GitHub issue, blog post).
• Have a track record of working async with people in other time zones.
• Are comfortable in GCP (Cloud Run, Cloud SQL, Cloud Tasks) — or AWS with a clear willingness to ramp.
• Push back when something looks wrong — including code written by cofounders, by other engineers, or by AI. We value disagreement that's grounded and direct.
  

• You've built features that use LLM APIs in production — function calling, tool use, RAG, agent loops, structured generation.
• You've worked in vertical SaaS — construction, field service, logistics, fleet management, real estate ops, or anything else where "the customer is in the field" was the operating model.
• Experience with WebSockets, Cloud Tasks (or similar job queues), transcription pipelines, or document processing (PDF/Excel generation, OCR).
• Open-source contributions, side projects, or prior founder experience. You build things even when nobody's paying you to.
• Bilingual English / Spanish or English / Portuguese.
• Have used PostHog for product analytics or Sentry for production triage in a serious way.

• Contract engagement structured as an independent contractor relationship (1099-equivalent local structure; we work with several common LATAM EOR/contractor models).
• Competitive contractor rate, calibrated to your country and experience level.
• AI-native dev environment. We pay for Claude Code, Cursor, and any tooling that makes you faster.
• Path to extended or expanded engagement based on fit at the end of the initial term.
  

• W-2 employment, US benefits, or equity at this stage. This is a contract role.
• A heavy meeting culture or a PM-driven roadmap. You're expected to think and decide.
• A pristine greenfield codebase. You'll inherit a real production system.