Full-Stack Product Engineer — LATAM Remote, US Hours

Carbon Technology · BeeSuite · Contract-to-Hire

Carbon Technology builds BeeSuite: the AI operating system for service franchises. Our platform recruits their employees, answers their customers’ calls, retains their workforce, and automates the operational work that usually falls between disconnected systems and overloaded teams.

This is not a demo. Our voice agents have answered over 9,000 production calls; our virtual recruiter has held more than 5,000 AI voice interviews; real franchise locations run their daily operations on what we ship. We have customers waiting to onboard, and one constraint on growth: how fast a deliberately tiny engineering team can build.

We need a full-stack product engineer—a strong mid-level ready for real ownership—who takes a customer problem from “this dashboard number looks wrong” all the way to designed, built, deployed, and verified in production. Not a ticket-taker. Not support with an engineering title.

How we build

We’re real coders—and we’re AI-first. We were shipping software long before AI wrote a line of code, but we embraced it early, and now one engineer here ships like five. We use AI aggressively at every layer: agents in production, agents in our own tooling, and AI in how we write, test, and review code.

If your instinct on a boring problem is “I’ll build an agent so nobody ever has to do this again,” you’ll feel at home. If your instinct is to paste AI output into production and hope, you won’t—architecture, security, and judgment still belong to the engineer.

We stay close to the problem. The engineer who builds a feature knows who uses it and what success means. The person who debugs a broken workflow is the person who heard the complaint.

We ship continuously. Small, verified improvements beat big releases that disappear into planning. No committees between you and production.

What you’ll own

Features, problem to production. Take a customer or product problem from an unclear starting point through design, implementation, testing, deployment, and verification against real production behavior—UI, API, database, and the integrations around them. You’ll have experienced engineers to collaborate with, but a well-understood problem shouldn’t need someone to specify every step.

Production problems that begin as customer questions. A report arrives as “the candidate never got the message” or “the system moved someone to the wrong location.” You turn that into an engineering answer: reproduce it, trace it across the frontend, API, database, background workflows, and external integrations, separate defect from configuration from misunderstanding, fix the cause, and add the test or monitoring that keeps it from returning silently. A patch is not done until the customer’s problem is actually gone.

The customer loop. Our account team owns the relationship, but you’ll join customer conversations when direct technical discovery helps—sitting with a nontechnical operations manager, working out what’s actually wrong underneath what she’s describing, and translating it into an engineering plan. Then explaining what happened back to the account team in plain language.

A product that has to earn trust. BeeSuite is multi-tenant software handling real operational and applicant data. Tenant isolation, authentication, input validation, migrations, observability, and testing are the job—we move fast, but speed never means gambling with customer data.

Work you might do here

• Build a multi-step recruiting workflow across React, APIs, PostgreSQL, and an external integration
• Trace why an automated SMS succeeded for every location but one
• Add an administrative surface without exposing one customer’s data to another
• Investigate a “wrong” dashboard number by tracing its complete data lineage
• Improve an AI-powered workflow while preserving the deterministic guardrails around it—or build a new agent that removes a category of manual work entirely

Our stack

TypeScript monorepo: Next.js App Router + React · Node.js · Supabase/PostgreSQL with Row Level Security and multi-tenant data models · NextAuth · Tailwind · Vitest + Playwright · Turborepo · Vercel.

Around the core: n8n, Twilio, LiveKit, Retell, and models from Anthropic, OpenAI, and Google.

You don’t need every keyword. We care about strong fundamentals and how quickly you understand a system you’ve never seen.

Who you are

You have roughly 3–6 years of professional experience shipping production web applications across more than one layer of the stack, and you can demonstrate:

• Strong TypeScript/JavaScript fundamentals and modern React experience
• Backend and API development, relational database design, and real SQL
• Genuine computer-science depth—data structures, system design, and failure modes—not just framework fluency. No degree required, but the understanding is; our interviews probe it.
• Systematic production debugging and the habit of verifying in production, not just in tests
• Daily fluency with modern AI development tools—without outsourcing your engineering judgment to them
• The judgment to choose the simple solution without deleting the safeguards

Two hard requirements—please self-select honestly

• Fluent spoken and written English. Every interview is conducted in English, and you’ll speak live with US customers and teammates. “Comfortable reading documentation” is not the bar.
• A full workday on US business hours. Your regular workday must align with an agreed US schedule—typically 9:00 a.m.–5:00 p.m. Pacific or Eastern Time. Pacific alignment is preferred. This is a full-day schedule, not a few hours of overlap.

Especially strong if

• You’ve owned features from customer discovery through production at a small B2B SaaS company
• You’ve built multi-tenant applications or shipped with Next.js, PostgreSQL, Supabase, or Vercel
• You’ve integrated external APIs with inconsistent or imperfect data
• You’ve built agents or AI-powered systems that run unattended in production
• You’ve been a technical founder, solutions engineer, or implementation engineer who still wrote production code

This role is not right for you if

• You want every task fully specified before you begin
• You want to work only on the frontend or only on the backend
• You consider production support someone else’s responsibility
• You avoid conversations with customers or nontechnical teammates
• You treat AI-generated code as correct until someone else proves otherwise

Structure and compensation

We’re not asking you to leap and hope.

Phase 1—three-month paid contract. Full-time, real product work from week one. $5,500–$7,500 USD in gross monthly cash compensation, based on experience and demonstrated ability. Either side can end it—no hard feelings.

Phase 2—long-term conversion on mutual fit. $66,000–$90,000 USD in gross annual cash compensation, meaningful equity as an early engineering-team member, and a path toward senior IC or technical leadership.

Depending on your country and mutual preference, the long-term relationship may be structured as employment through an Employer of Record or as a long-term international contractor. EOR employment will include the locally required employment benefits. Contractor compensation and any additional benefits will be agreed directly.

An exceptional candidate who needs a different structure should still apply—the structure serves the fit, not the other way around.

Deliberately funded. Founders with prior exits and ex-Apple leadership. We raise carefully and only what the mission needs, so the people building this keep real ownership and a long runway.

Hiring process

You apply through RecruitBee—the same hiring platform we build for our customers. Experiencing the product from the candidate’s side is part of the process.

1. A quick call with one of our RecruitBee AI agents. You’ll answer five or six introductory questions.

2. Meet with the team. We’ll have a technical and product discussion. Expect real computer-science and system-design questions, conducted in English.

3. A customer-communication conversation. We’ll evaluate how you understand an ambiguous problem, communicate with nontechnical stakeholders, and translate what you learn into sound engineering decisions.

4. References and decision.

We’ll ask you to walk through something you owned across the stack—the architecture, the tradeoffs, what went wrong, and how you proved it worked. And we’ll ask about an ambiguous production problem: how you investigated, how you communicated while you didn’t yet know the answer, and what you changed afterward.

If that’s the kind of engineering you want to do, we’d like to meet you.