Backend Engineer

Most teams are still debating whether AI should write production code. We are past that debate: Claude Code is how we build, every day, across all our products. What we need now are senior engineers who make that fast and correct — and who enjoy being trusted with real responsibility.

alpview it service is a small software company building long-term products for established European companies, mainly in the HR and staffing domain: Laravel backends, Angular portals and Flutter apps, with real users, real contracts and real money moving through them.

WHAT YOU GET

- A long-term contract that works like employment: paid vacation, paid public holidays, a steady full-time workload and reliable monthly payment. No gig hopping, no "we'll see after the sprint"
- Claude Code as your daily tool, with the licence paid by us — in a team that actually uses it well, not one that just wrote a policy about it
- Real ownership: you will own parts of the system, and your technical arguments change how we build. We would rather be convinced than obeyed
- Short decision paths: you talk directly to the people who decide, not through three layers
- Work that matters: production systems where correctness counts — documents, contracts, integrations, money. Interesting problems, not CRUD for its own sake
- Products built to last for years, and more work coming
- Full remote, flexible hours, async-friendly, very few meetings
- A fast, respectful hiring process — no leetcode, no unpaid test assignment (details below)

WHAT YOU WILL WORK ON

You work on our Laravel backends — the core of several products, used by mobile apps and web portals that other developers own. The systems are live or close to it.

The work is a mix of new features, integrations with external systems, and hardening what is already there so that it behaves exactly the way we say it does. There is a lot of meaningful work on the table right now, with fixed client deadlines. For you that means: no waiting for a project to start and no long onboarding queue. After a short introduction you take on real stories and own them end to end — from understanding the requirement to a reviewed, tested and deployed result.

This suits engineers who find their way around an existing codebase quickly, ask the right questions early, and like being handed something important rather than a detailed instruction for every step.

We keep client specifics out of a public posting. You get the full picture, including a walk through the real codebase, in the first conversation.

HOW WE USE CLAUDE CODE

Our pace is genuinely higher than it was a year ago. That raised the bar on the human side, and that is exactly the interesting part of this job: you direct the agent. You review its code the way you would review a junior's PR, you notice when it adds a second service that does what an existing one already does, when a "defensive" try/catch silently swallows the error that mattered, or when a test asserts nothing real — and you say no when saying yes would be faster.

One rule above all: you are responsible for your code. Every line you commit is yours, no matter who or what typed it. "The AI wrote it" is never an explanation for a bug, a security hole or a broken migration. You understand every change before it goes into a pull request, you can explain and defend it in review, and you validate what the agent produced — by reading the diff, running and extending the tests, and checking the behaviour yourself, including the failure paths. The agent makes you faster; it does not take the responsibility off your shoulders.

Good Claude skills are half the job here. If this is how you already work, you will feel at home immediately:

- CLAUDE.md files and project memory you maintain, so the agent knows the house rules
- Your own slash commands, skills and hooks for recurring work
- MCP servers, subagents, worktrees for parallel work, sensible task slicing
- Knowing when to give the agent a long leash and when to keep it to one file
- Handing work over cleanly between sessions so nothing is lost when a context ends

If you use agents well but differently, we want to learn from you. Several developers work on the same codebases in parallel, so we value consistency over cleverness.

WHAT THE CODEBASE IS LIKE

Not a greenfield, and not a rescue mission either.

- Modern Laravel on current PHP, with a real test suite (feature and unit) that runs in CI on every pull request
- MySQL and Redis, queues and scheduled jobs, everything in Docker
- Asynchronous work that has to be right: retries, idempotency, unique jobs, recovery commands for what got stuck. "The job ran twice" is a bug we take seriously
- Webhooks in and out, integrations with external services and enterprise systems
- Document generation and document workflows, with an audit trail
- Multi-tenancy and multiple languages
- Architecture decisions written down as ADRs
- REST APIs consumed by Angular and Flutter clients — so API contracts are discussed, not just changed
- A short written evidence record per story: commit, environment, steps, expected vs. actual, including the negative paths. It is the reason acceptance rounds go smoothly

THE DEAL, IN BOTH DIRECTIONS

Formally this is a long-term freelance contract, as with all our developers in Ukraine. In practice we work together like a good employment: paid vacation, paid public holidays, a steady workload, and a cooperation meant to last for years. You are a member of the team, not an external resource.

We ask for the same commitment back:

- This is your main engagement, not one of several projects you juggle. If something else starts competing for your time, tell us first
- Transparent working hours: you track your time in a way we can follow, and you say early when you are behind or unavailable. That is what makes paid vacation and flexible hours possible
- Ownership when it counts: occasionally, when a release is in flight or something breaks in production, that can mean an evening. It is the exception, and those hours come back to you as time off, whenever it suits you

WHAT YOU BRING

- 6+ years of professional PHP, of which at least 3 with Laravel in production
- Real production experience with Claude Code or a comparable agent workflow — daily work where the agent writes a large share of the code and you validate and own the result
- Solid architecture and review instincts: service boundaries, testability, and the discipline to match an existing pattern instead of inventing a better one
- Queues, jobs and scheduling in depth: retries, timeouts, idempotency, failure handling
- MySQL beyond CRUD: migrations on live data, indexes, transactions, locking
- Testing with PHPUnit or Pest that tests behaviour, not implementation
- Docker, CI pipelines, pull requests and code review as normal daily practice
- Careful handling of secrets and personal data (GDPR is a real constraint here, not a footnote)
- You become productive quickly in an unfamiliar codebase and take responsibility without being micromanaged
- Available to start soon
- English B2+, and the willingness to argue a technical trade-off rather than only execute a ticket

Nice to have: German; HR, staffing or ERP domain experience; e-signature or document workflows; enough Angular or Flutter to read the client side; your own MCP servers or agent tooling.

REQUIRED SKILLS EXPERIENCE

- PHP — 6 years
- Laravel — 3 years
- MySQL — 4 years
- Claude Code / AI-assisted development — 1 year

Required languages: English — B2 (Upper Intermediate)