Planning: I start with a short spec — roles, date filters, grouping, export format, and expected row volume. I prompt the agent with the spec, stack (Laravel, PostgreSQL, students/classes/attendance tables), and ask for edge cases: absent vs late, timezone boundaries, soft deletes, and authorization.

Implementation: AI scaffolds migrations and index ideas, Eloquent models with eager loading, Form Request validation, Policy stubs, and a test list. I own the query layer — I fix N+1 and joins myself. AI drafts the controller, API Resource, and PHPUnit/Pest stubs; I edit every line before merge.

Review: Feature tests cover auth, empty ranges, and date boundaries. I run EXPLAIN on the main report query. I do not merge unreviewed SQL. Security and data integrity are non-delegable — same discipline I applied on ayo.de with AI-assisted delivery.

Deploy: CI green → pipeline deploy. For heavy reports I use a feature flag or async export in v1. I document the report definition for support. AI shortened boilerplate; I own production.

The smallest production slice where users complete one valuable workflow end-to-end with reliability and observability — not a demo.

It forces one problem, one metric, and one release path. My wins at TLSContact and in consulting came from narrow slices with rollback plans and monitoring.

MVP means explicit now / later / measure — what ships this week, what waits, and how we know it worked.

At TLSContact during microservices migration we cut by domain — new work shipped as services, legacy stayed behind stable APIs.

I cut polish before auth, auditability, or data correctness. Every cut is documented with a reason for stakeholders.

At Aubay I delivered thin vertical slices approved in writing before build — so scope debates happen upfront, not on launch day.

Yes — when the critical path is tested and monitored. I name gaps explicitly in the PR or handover and ticket follow-ups.

I never ship known security or data integrity bugs.

At INBOX we shipped monitoring iteratively — alerts first, then dashboards — so teams got value before the suite was perfect.

I assume good intent, then specify the broken outcome, operational cost, and trade-offs. I offer alternatives with effort and risk estimates.

I disagree with data in the open — latency numbers, support load, migration cost.

If overruled, I commit, document constraints, and add observability so we learn fast. I make trade-offs visible; I do not need to win every argument.

At Arcocia a third-party API slip threatened a deadline. I escalated before the due date with status → impact → options → recommendation. The team chose a phased delivery instead of a silent slip.

At Proxify I re-scoped in writing when requirements shifted mid-sprint — new acceptance criteria before more code.

Early flags build trust; late surprises destroy it.

Peak load was TLSContact full-time plus Aubay consulting (5+ clients). I use WIP limits, written scope per engagement, batched deep-work blocks, living notes per project, and a shared definition of done.

AI helps context re-entry — summaries, test drafts, doc stubs, diff review — but I do not delegate merge ownership.

Each engagement gets a one-page scope doc updated when priorities shift.

Calm under pressure and a concrete communicator — I prefer specifics over vague optimism.

At TLSContact colleagues pointed to performance ownership and monitoring that lasted after I moved on. At INBOX: clear stand-ups and maintainable handoffs.

My natural role is deep on hard backend problems — queries, service boundaries, releases — and closer on scoped, tested delivery.

I am open to discussing compensation once we align on role scope, seniority, employment type (full-time vs contract), location/time zone, and benefits.

I am flexible between USD and EUR arrangements for the right fit and care more about scope, team, and impact than a number in isolation.

Happy to share a range in a direct conversation after those details are clear.