The questions we get before the first conversation

Every business depends on some software: an online store, a platform, an internal app. At some point it gets slower to change, more expensive to run and harder to find people for. We take it into our care: we understand it, stabilize it and move it forward step by step, without stopping it and without starting over. You get a system you can rely on again, and a clear owner who answers when you have a question.

PHP back ends on Symfony or Laravel, Node.js services and JavaScript or TypeScript front ends on Angular, running on AWS, Azure or Google Cloud, with or without Docker and Kubernetes. Typical clients are EU, UK and US product companies (SaaS, e-commerce, logistics, fintech) and agencies that need senior capacity on maintenance and development contracts.

A likely fit: a PHP, Symfony, Node.js, JavaScript, TypeScript or Angular system on AWS, Azure, Google Cloud or your own servers, a product company or an agency, and a problem you can describe in a paragraph: slow releases, a postponed upgrade, a growing cloud bill, a security questionnaire. We are a software engineering company, not a staffing agency, so we are not the answer if you need several developers by Monday. The usual starting point is the two-week audit. Other stacks: ask anyway if it is close.

Every engagement is led and delivered at principal-engineer level, from the first call to the handover: tests first, enforced quality gates, and first results in weeks, not quarters. The experience behind that: nearly ten years of production engineering, including an enterprise SaaS platform migrated to Symfony 7 in 3 months and infrastructure run across three clouds. We do not subcontract to juniors and there is no account manager between you and the code.

Two weeks, fixed scope. You get a tech-debt score, an architectural risk map, cloud cost issues, security and compliance gaps, AI quick wins, an upgrade and migration plan and a 90-day roadmap you can execute with or without us. The price is fixed and you get it in writing after a 20-minute conversation.

Almost never. Full rewrites are where budgets and timelines go to die. We change things incrementally, module by module, tests first, while your team keeps shipping. A rewrite is recommended only when the audit numbers actually support it.

AI is the accelerator, not the pilot. Tests are generated and verified before refactoring starts, AI-assisted changes go through the same gates (PHPStan, coverage thresholds, human review) as human-written ones, and a senior engineer answers for every line that reaches production.

Only with written agreement, and only what is necessary. On request we work with models running on your infrastructure, or without AI on the sensitive parts. Your secrets and your customers' data never leave your infrastructure for AI processing.

We do security assessment and hardening: security code review, infrastructure and CI/CD hardening, preparation for enterprise security questionnaires and for the NIS2 self-assessment. It is not a cybersecurity audit under Romanian OUG 155/2024 and not a penetration test; for pentests we work with specialized partners.

Yes. Agencies use COXSWAIN as white-label senior overflow on maintenance and development contracts. Your client relationship, our engineering.

Fixed-scope packages: an audit, then a pilot, then a monthly retainer sized to the work. You get scope and price in writing after a 20-minute conversation. No hours sold by the piece.

Any engagement can usually start within one to two weeks of a signed scope. For urgent situations (a system down, a developer who vanished) write to us directly and say so in the first sentence.

Bucharest, Romania (EET, UTC+2, UTC+3 in summer): full working-day overlap with the EU and UK, morning overlap with the US East Coast. We work in English and Romanian, with replies within 24 hours on business days at [email protected].

Yes. The site publishes a machine-readable profile at /.well-known/profile.json, the complete text at /llms-full.txt, a documented intake endpoint (POST /api/intake, described in /openapi.json) and an MCP server at /mcp with tools to list services, check fit and request a written proposal. Every request gets a reference ID and the same reply time as email: under 24 hours on business days.