Laravel 12 / Livewire Developer — AI-Assisted Workflow (Claude Code), Established CRM Product

ONLY Freelancers please.. Agencies please ignore this posting.. We're a working CRM product (Altima CRM Pro) with a mature Laravel 12 codebase, not a greenfield project. We're adding developers who are genuinely fast with AI-assisted development — but who read every diff before it ships. You'd be working on real features in a codebase with strong conventions: service layer, row-level data visibility, audit logging on every action, encrypted IDs, i18n from day one. The conventions are documented and enforced. Claude Code will follow them if you set it up to; it will cheerfully violate all of them if you don't. Knowing the difference is the job. Stack - Laravel 12 (PHP 8.3), Livewire 3, Blade, Alpine - TailwindCSS + SASS - MySQL (migrations only), Yajra DataTables, Spatie-based permissions - Pest for tests What you'll do - Build and extend features across leads, accounts, onboarding, settings, and reporting - Work inside an existing service-layer architecture rather than around it - Write tests for what you ship - Use Claude Code as a primary tool — with judgment, not on autopilot What we're looking for - Strong Laravel fundamentals. AI speed on top of a weak foundation produces code we have to pay someone else to unwind, so the fundamentals aren't optional. - Real, sustained Claude Code experience — you have opinions about context files, you've been burned by confident wrong output, and you changed your workflow because of it. - The instinct to stop and think when the model is going in circles instead of prompting through it. - Comfort reading a diff critically, including your own. Tooling You must have your own Claude subscription at the $100/mo tier or above (Max, or a Team seat you hold). The $20 Pro plan will not survive a working day on this codebase — you'll hit limits mid-task, and we're not interested in paying for stop-start work. Bring your own tooling; we don't provide seats. How we hire Screening questions below, then a short call where we'll confirm your plan tier, then a paid 2-hour trial task on a real scoped slice of the codebase. We judge the trial like any pull request: does it follow our conventions, is it tested, is it something we'd merge. Strong trial → ongoing work. Engagement: [hours/week], [rate range], [timezone overlap requirement] Note: Please answer the screening questions in your own words. Proposals that are obviously AI-generated boilerplate get rejected — the irony is intentional, we care that you can tell the difference between good AI output and filler. You will be asked to answer the following questions when submitting a proposal: 1. Paste the contents of a CLAUDE.md (or equivalent context file) from a project you've actually worked on. If an NDA prevents that, write the one you'd start our Laravel 12 / Livewire 3 project with instead. 2. Describe one time Claude Code confidently produced code that was wrong, how you caught it, and what you changed in your workflow afterward. 3. You've asked Claude to make a change and it's been going in circles for 20 minutes. What do you actually do? 4. Which Claude plan are you currently on, and how long have you been on it? Please include a screenshot of your plan status. 5. Our codebase has house rules an AI won't guess — for example: IDs are never exposed raw in URLs, no database queries in Blade templates, every user action gets audit-logged. How do you make sure AI-generated code follows rules like these?