Staff Software Engineer , Product

LawnStarter is the nation's leading on-demand marketplace for lawn care and outdoor services, with over $100M in annual bookings.

Local

Belo Horizonte - MG

Remoto

Responsabilidades

• You have a hand in the full lifecycle: shaping the problem, deciding the technical approach, directing AI agents to implement much of the code, shipping to production, and — with your team — owning the outcome.
• You're measured by impact, not by lines of code merged.
• When an agent can ship something safely, your job is to make sure it's done right and the metric moves.
• When the work calls for careful, hand-written code in a sensitive area, you write it yourself.
• You work as a true product partner.
• You sit at the table with PM and design, bringing engineering judgment to product calls and product sense to engineering calls.
• You get real autonomy — with the right checkpoints.
• You make most technical calls yourself, with architect review on significant architectural decisions and fast input from peers.
• You operate at a staff bar.
• You're trusted to make the call, ship the hard thing, and stand behind the outcome.

Requisitos

• AI-native.
• Claude Code, Cursor, Codex, or equivalent are how you ship today — daily, on production work.
• You have real opinions about prompts, evals, agent loops, and review workflows, and you know when to let the agent run versus write it yourself.
• Operating at a lead level.
• Outcome-driven.
• You measure your week in "did the metric move" and "did the experience get better." You read the post-launch dashboard and own the answer.
• A strong horizontal partner.
• You hold your own with a strong PM and designer, and you collaborate well with engineering peers in a shared codebase.
• You bring engineering judgment to product calls and product judgment to engineering calls.
• Decisive and documented.
• You make architecture, data-model, and rollout calls, write them down, get fast input, and move.
• A force multiplier.
• Your impact compounds beyond your own initiative because you leave reusable artifacts behind — agent workflows, evals, runbooks, post-launch reviews.
• Customer- and pro-minded.
• This is a real marketplace with real people on both sides, and you care about the outcomes for both.

Diferenciais

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

Sobre a empresa

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 build in small, focused initiative teams: a Product Engineer working alongside a PM and a designer, supported by an Engineering Manager who helps you grow.

You'll also work shoulder-to-shoulder with engineering peers across initiatives in a shared codebase.

The whole team owns whether the work moves its metric.

AI coding agents are a force multiplier here — they give a small, senior team the leverage to ship more, faster, and at a higher bar for quality.

We hire engineers who are wired for ownership and energized by shipping to a real marketplace with customers and pros on both sides.

The Role You're the engineering anchor of an initiative — working as part of a tight team with your PM and designer, and alongside engineering peers on adjacent initiatives.

What makes this role exciting: You ship end-to-end.

From problem-framing through production to the post-launch metric review — you see the whole arc and own the result with your team.

What You'll Own The technical approach — architecture, data model, integration choices, rollout plan, observability, and rollback strategy for your initiative.

You make most calls yourself and bring significant architectural decisions to architect review; you document them, and revisit if the data says you were wrong.

Implementation quality — the prompts, guardrails, evals, tests, and review loop that let agents ship safe, correct, production-ready code.

Most lines will be agent-authored, and you're accountable for them — held to the same standard as the rest of the team working in a shared codebase.

Cross-functional partnership — daily working contact with your PM (scope, tradeoffs) and designer (UX decisions, in-tool prototyping), regular collaboration with engineering peers, and weekly check-ins with your EM.

The initiative outcome — the metric the initiative was set up to move.

With your PM, you present results 2–4 weeks post-launch and share the "did it work" answer.

A high bar for what ships — production correctness, security, 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 a staff-level quality bar Most of the code on your initiative will be authored by AI agents.

The craft is making them ship as if a senior engineer wrote it: prompts that encode our conventions, evals that catch issues before merge, tests that exercise the edges, observability that catches a regression before a customer does.

How do you build a workflow that lets a small team ship far more than its size would suggest?

Owning decisions with high autonomy You have real latitude to make and document technical calls quickly — with architect review on the big architectural ones and peers to pressure-test your thinking.

How do you move fast, keep your team aligned, and stay accountable to the outcome?

Shipping outcomes, not features Each initiative is measured by a metric — a conversion rate, a retention curve, a pro-funnel KPI, a unit-economics shift.

You're accountable for the number alongside your team.

How do you scope to actually move it, decide what *not* to build, and have the discipline to follow up 2–4 weeks after launch?

What Success Looks Like (Year 1) Initiative outcomes hit — You've shipped 3–4 initiatives end-to-end, and at least two clearly moved their metric (with the post-launch review to prove it).

Agent workflow that travels — The prompts, evals, and review loop you built are picked up by peers on other initiatives.

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