[Hiring] Staff Product Engineer (São Paulo) @LawnStarter

This is a remote role for candidates located in São Paulo (Brazil)

About LawnStarter

LawnStarter is the nation’s leading on-demand marketplace for lawn care and outdoor services, with over $100M in annual bookings. We’re expanding beyond lawn care to become the one-stop shop for all home services — operating across three brands (LawnStarter, Lawn Love, Home Gnome) on a single shared platform.

About Engineering at LawnStarter

We’re restructuring engineering around initiative teams: a Product Engineer paired with a PM and a designer, with an Engineering Manager who covers a couple of initiatives and supports your growth. The engineer leads AI agents like a team, ships the work, and is accountable — with the rest of the triangle — for whether the initiative moves its metric.

We’re betting that 1–2 strong engineers running AI agents can outship the labor-team model that defined the last decade of software. That bet only works if the engineers we hire are wired for ownership and can ship to a marketplace with real customers and pros on both sides.

The Role

You’re the engineering anchor of one initiative at a time. The initiative is a team effort — an iron triangle of you, your PM, and your designer — and you have key participation across the full lifecycle: shaping the problem, deciding the technical approach, leading the AI agents that implement most of the code, shipping to production, and answering for the outcome alongside the rest of the triangle.

You’re accountable for the outcome — not for the volume of code merged. If an agent can ship it safely, your job is to make sure the agent does it right and the metric moves. If the initiative needs hand-written code in a sensitive area, you write it yourself.

What makes this role different:

• You lead AI agents, not humans. Claude Code, Cursor, Codex, and our internal agent stack are your team. You own the quality, safety, and velocity of what they produce.
• You own an outcome, not a ticket queue. Problem-framing through production through the metric review 2–4 weeks after launch.
• You partner horizontally with PM and design. No tech lead above you. No architect approval. No ticket grooming committee.
• The bar is staff, not senior. You make the call when the call needs to be made. If you’re waiting to be told, this isn’t the role.

What You’ll Own

• The technical approach — architecture, data model, integration choices, rollout plan, observability, and rollback strategy for your initiative. You make the call, document it, and revisit it if the data says you were wrong.
• Agent-led implementation quality — the prompts, guardrails, evals, tests, and review loop that let agents ship safe, correct, production-ready code on your initiative. Most lines will be agent-authored. You’re accountable for them.
• Cross-functional partnership — daily working contact with your PM (scope, tradeoffs) and your designer (UX decisions, in-tool prototyping with agents), and weekly check-ins with your EM (initiative health, blockers, growth).
• The initiative outcome — the specific metric the initiative was set up to move. In partnership with your PM, you present results 2–4 weeks post-launch and share the “did it work” answer.
• A high bar for what ships under your name — production correctness, security posture, performance, observability, and the experience for customers and pros. Agents accelerate you; they don’t lower the bar.

Problems to Solve

Leading AI agents at staff-level quality
Most of the code on your initiative will be authored by AI agents. The work is making agents ship as if a senior engineer wrote it: prompts that encode our codebase conventions, evals that catch hallucinations before merge, tests that exercise the edges, observability that catches the regression in production before a customer reports it. How do you build the agent workflow that lets one engineer ship what used to take a team?

Owning an outcome without a tech lead
You don’t have a tech lead to approve your design or an architect to escalate to. You have an EM who covers a couple of initiatives and peers on adjacent ones. How do you make calls fast, document them clearly, and stay accountable to the outcome — without slowing down for hierarchy that no longer exists?

Shipping outcomes, not features
The initiative will be measured by a metric — a conversion rate, a retention curve, a pro-funnel KPI, a unit economics shift. You’re accountable for the number, not the feature. How do you scope to actually move it, decide what to not build, and have the discipline to follow up 2–4 weeks after launch even when the next initiative is calling?

What Success Looks Like (Year 1)

• Initiative outcomes hit — You’ve shipped 3–4 initiatives end-to-end, and at least two clearly moved the metric they were set up to move (with the post-launch review to prove it).
• Agent workflow that travels — The prompts, evals, and review loop you built for your initiative are adopted by at least one other engineer on an adjacent initiative.
• Cycle time — Median time from problem-framing to first production rollout on your initiatives is meaningfully shorter than the pre-restructure baseline.
• Zero “agent-shipped that” incidents — No customer- or pro-facing regression traceable to agent-authored code that you missed in review.
• Visible leverage — Other engineers point to artifacts you left behind — runbooks, evals, agent workflows, post-launch write-ups — as references they use.

Who You Are

AI-native. Claude Code, Cursor, Codex, or equivalent are how you ship — daily, on production work. You have opinions about prompts, evals, agent loops, MCP servers, and review workflows, and you know when to let the agent run vs. write it yourself. This is unlikely to be a good fit if you describe AI coding as “something you’re exploring” or prefer to write everything by hand.

Already operating at lead level. You may currently be titled Senior, Staff, Lead, or Principal — but in practice you’ve been the person making the call, shipping the hard thing, and answering for whether it worked. This is unlikely to be a good fit if you’ve always had a tech lead breaking down the work for you.

Outcome-driven, not output-driven. You measure your week in “did the metric move” and “did the experience get better,” not in tickets closed. You read the post-launch dashboard and you own the answer. This is unlikely to be a good fit if you take pride in volume of code shipped or feel uncomfortable being measured on a number you don’t fully control.

A strong horizontal partner. You hold your own with a strong PM and a strong designer. You bring engineering judgment to product calls and product judgment to engineering calls. This is unlikely to be a good fit if you hide behind “that’s product’s decision” or default to RICE-scoring tickets handed down to you.

Decisive and documented. Architecture decisions, data-model choices, rollout plans — you write them down, get fast input, and move. This is unlikely to be a good fit if you wait for consensus on questions that have a clear right answer, or if you make calls and never write them down.

Raises the floor, not just the ceiling. Your impact compounds beyond your own initiative because you leave artifacts — agent workflows, evals, runbooks, post-launch reviews. This is unlikely to be a good fit if you’re a lone wolf who ships brilliantly but leaves nothing reusable behind.

Cares about customers and pros. This is a real-world marketplace with real people on both sides. This is unlikely to be a good fit if you’re chasing pure engineering elegance over business and customer outcomes.

This Role Is NOT

• A tech lead in an old-style team. No 4–5 engineers reporting up to you on technical direction. The team is you + PM + designer + EM, with AI agents doing most of the implementation.
• A management role today. People management is the EM’s job in this role. That said, the path can grow into management for those who want it — it’s an open door, not a closed one.
• A platform-only or architecture-only role. You’re a Product Engineer. You ship features that move metrics, end-to-end. Platform work happens inside the initiative when it’s needed for the outcome.
• A “let AI do everything” role. Agents handle implementation grunt work. You handle judgment, design, safety, and accountability. The bar is higher than the old senior bar, not lower.
• A research role. This is shipping to a marketplace with $100M+ in bookings. Customers and pros are using what you ship inside the same week.

Tech You’ll Touch

• AI agents — Claude Code, Cursor, Codex, internal agent stack, MCP servers, evals tooling
• Backend — PHP/Laravel
• Frontend — TypeScript/React/React Native (customer & pro apps, web and mobile)
• Data — Redshift, dbt, Segment, Airflow
• Infra — AWS, Datadog, Sentry, GitHub Actions
• Documentation & process — Brain (Claude Code skills + docs repo), Confluence, Jira

You don’t need every box checked. You need deep skill in at least one of our stacks plus credible production experience with AI coding agents.

Benefits

• Competitive salary of USD $80,000–$100,000 annual base
• Work from anywhere
• High ownership and autonomy
• Fast-moving team that loves to build, learn, and grow