Counterpart is an insurance company that builds an operating system for business risk and develops Agentic Insurance combining insurance expertise with AI for management and professional liability solutions.
<h1><strong>Staff PRODUCT ENGINEER (Contract)</strong></h1> <p><em>Build it. Ship it. Own it.</em></p> <p><strong>Location:</strong> Argentina, Brazil, Colombia, Chile, Mexico, Dominican Republic, Peru</p> <h2><strong>About Counterpart</strong></h2> <p>Counterpart is an insurtech platform reimagining management and professional liability for the modern workplace. We believe that when businesses lead with clarity and confidence, they become more resilient, more innovative, and better prepared for what's ahead. That's why we built the first Agentic Insurance™ system — where advanced AI and deep insurance expertise come together to proactively assess, mitigate, and manage risk. Backed by A-rated carriers and trusted by brokers nationwide, our platform helps small businesses grow with confidence. Join us in shaping a smarter future, helping businesses Do More With Less Risk®.</p> <h2><strong>The Role</strong></h2> <p>As a Product Engineer at Counterpart, you own problems and drive outcomes. You iron out requirements with stakeholders, partners, and customers. You leverage AI for implementation. You verify the end-to-end result. You iterate until you move the needle.</p> <p>You run projects with other Product Engineers and AI agents. You know when to delegate to a human, when to delegate to an agent, and when to do it yourself. </p> <h2><strong>YOU WILL</strong></h2> <h3><strong>Own the Problem</strong></h3> <ul> <li>Take accountability for the outcome. </li> <li>Talk directly to your users, stakeholders, and peers to discover the problem. </li> <li>Translate business intents into functional requirements that engineers and agents can build against. Determine where human judgment is non-negotiable and where AI can augment or act autonomously. </li> <li>Own the non-functional technical requirements. Specify the quality bar: performance, scalability, reliability, security, maintainability.</li> </ul> <h3><strong>Build with AI</strong></h3> <ul> <li>Write production-quality code across the stack: Python, Django, AWS, React, and PostgreSQL. You are not afraid of any layer.</li> <li>Direct agents and review their output with domain depth. Catch what they miss across the functional and non-functional bar: correctness, edge cases, performance, security, testability, readability.</li> <li>Evaluate and improve agent output regularly. Track failure modes. Refine the prompts and patterns that lift the quality ceiling over time.</li> <li>Run projects that span other Product Engineers and agents. Coordinate work, manage dependencies, and keep things moving without micromanaging.</li> <li>Contribute to architectural decisions. Bring options and alternatives, not objections. Know when the elegant solution is the wrong call for a startup moving fast.</li> <li>Contribute to and manage the context harness as you build and deliver.</li> </ul> <h3><strong>Verify and Iterate</strong></h3> <ul> <li>Surface delivery risks early. Bring solutions, not just problems. Keep your stakeholders informed without turning every risk into a meeting.</li> <li>Verify the end-to-end result yourself. Confirm the system does what the requirements say, and that the requirements were right.</li> <li>Drive adoption of what you ship. Documentation, enablement, and usage monitoring are part of the job.</li> <li>Monitor your services in production. Own the signal. Debug and fix what breaks.</li> <li>Participate in support rotations. When users run into issues with what you shipped, you investigate and fix them.</li> <li>Iterate until you move the needle. Shipping is not the finish line. Outcome is.</li> </ul> <h3><strong>Raise the Bar</strong></h3> <ul> <li>Mentor engineers around you. Give direct, useful feedback. Leave things better than you found them.</li> <li>Make deliberate calls on technical debt: when to accumulate it, when to pay it down, when to refactor. Translate trade-offs into language that business leaders can act on.</li> <li>Operate to the engineering standards for quality, reliability, and operational excellence.</li> </ul> <h2><strong>YOU HAVE</strong></h2> <ul> <li>10+ years building production web applications across the full stack. The years are the floor, not the bar. What matters is whether you have actually done this: owned a problem end-to-end, written the spec, built the thing, coordinated the work, and shipped it without someone else holding the thread.</li> <li>Demonstrated experience gathering requirements directly. You have interviewed users, customers, and partner teams, and turned ambiguous intent into specs that engineers and agents can build against.</li> <li>Demonstrated experience owning both functional and non-functional requirements. You have defined what a system must do and held the line on how well it must do it.</li> <li>A track record of owning a domain or initiative without close supervision. You have navigated ambiguity, made judgment calls, and been accountable for the outcome.</li> <li>Extreme ownership and a bias for action. You do not sit and wait. You chase stakeholders, customers, and answers when you need them.</li> <li>Experience coordinating across engineers and AI agents on a shared initiative. You have managed the work without holding a formal management title.</li> <li>Strong technical judgment and the depth to catch what AI-generated code gets wrong: not syntax, but correctness, architecture, intent.</li> <li>Proficiency with Python and Django. Working knowledge of TypeScript, React, and PostgreSQL.</li> <li>Proficiency in developing with AI tools (Claude Code, Cursor, OpenAI Codex, similar) and the drive to stay ahead of how AI development practices are evolving.</li> <li>Domain curiosity. You learn the business context your work supports, not just the systems. You understand why the problem matters, not just how to solve it.</li> <li>The ability to communicate clearly across engineering, product, and business, translating technical complexity without losing m