Senior Engineer / Architect — Independent Code & Architecture Reviewer (Laravel · Vue · PostgreSQL)

The role in one line We build fast with AI coding agents. We need an experienced human engineer whose job is to look at what's built and tell us, plainly, whether it's sound. About the product Clerro is a financial forecasting tool for small/medium sized businesses that make and sell physical products. We're pre-launch with one live test tenant, heading toward a self-serve public launch. The stack you'll be reviewing PHP 8 / Laravel, multi-tenant (schema-per-tenant on PostgreSQL 17), Redis Vue 3 + Inertia, an in-house component library. A Python calculation engine run as a queue worker. PHPUnit (~3,000 tests) and Puppeteer end-to-end walks that drive the real UI Laravel Forge, GitHub. What you'll actually do: Review each increment before it ships. We work in layers (setup → catalog → markets → pricing → demand → production → …). Each layer has a written design and a written definition of done. You read the code against both and tell us where it falls short. Judge the architecture, not just the diff. Is the data model right? Will this survive a hundred tenants? Is there one source of truth for each fact, or two that will one day disagree? Where will this be painful to change in a year? Hunt the class, not the instance. When you find a bug, say what family it belongs to and what test would make the family unrepresentable. Check the tests are real. A test that can't fail proves nothing. Part of the job is confirming the suite actually pins what it claims to. Write it down, in plain English. Findings go into a written report: verdict first, severity-ranked, file and line named, with what to do about it. The founder is not an engineer and will read your summary; the fix instructions will be handed to an AI agent, so they need to be precise. Occasionally, roll your sleeves up. Mostly you review. Sometimes the right thing is to fix it yourself or pair on a hard problem. That's welcome but not the default. You're a strong fit if you: Have 8+ years of professional backend experience, a good chunk of it in Laravel, and have owned the architecture of at least one system that ran in production with real users and real money. Are strong on relational modelling and SQL. You think about where a fact lives and who reads it before you think about UI. Know double-entry accounting or can pick it up fast — the engine enforces that journals net to zero and A = L + E, and you need to be able to tell whether it really does. Have reviewed other people's code for a living: as a tech lead, staff engineer, consultant or auditor. You know the difference between a style note and a finding. Are sceptical by nature. AI-generated code is fluent, plausible and confidently wrong in ways that don't show at the line level. You need to be the person who traces the fact through four files and notices it's computed a second time. Write clearly and directly. "This is wrong, here's why, here's the evidence" — not softened, not padded. Nice to have: Python Forecasting, FP&A, inventory or supply-chain domain experience Multi-tenant SaaS at scale Experience supervising or auditing AI-assisted codebases How we work: Everything is written down: a project constitution, an engineering watch-outs register, a design brief per page, a position document per area. You'll be given all of it. You'll be expected to read it and hold the code to it — and to tell us when the documents themselves are wrong. The founder makes decisions fast and wants them as short recommendation lists, not essays You'll be expected to disagree. If a design is signed off and it's wrong, say so with evidence. The product wins arguments, not the transcript. Async-first, written, with a regular call