The terms most vendors settle verbally, written down first.

Intake and scoping

The first call is 45 minutes and you spend it with an engineer, not a salesperson. Four things are asked on it: what the system is and what is wrong with it, who owns the backlog, who on your side will accept pull requests and settle priority, and which technologies the work sits in. If those fall outside the ten we staff against, you hear it on that call rather than in a proposal three weeks later.

Within three business days you get a scope note: what we think the work is, what we think it is not, the team shape we would put on it, and the two or three things most likely to go wrong. One or two pages, written to be argued with.

Nothing is staffed until that note is agreed in writing and turned into an SOW. Where the scope cannot be written down we say so and quote monthly work instead, because a fixed price against an undefined scope is one of the five things we decline.

Who you actually get

One or two named candidates per seat, with CVs, within five business days of a signed scope note. If we cannot put a specific engineer against your work we say the seat is not available, rather than send twenty CVs to sort through.

Every candidate is already on our payroll — we do not recruit against your order, so the person in the CV has been here long enough for us to know how they work. Median tenure is 2.8 years, median experience nine years. English is assessed against B2+ for engineers and C1 for team leads before anyone reaches you.

You interview each candidate yourself, with your own engineers, for as long as you want, in the technologies the work actually uses. You can reject anyone without giving a reason, as many times as it takes. That veto does not expire at the start date: if in week three the engineer is not what the interview suggested, tell us and we replace them under the rules below, at our cost.

Ramp-up

Ten to fifteen business days from signature to a productive engineer. Day one is accounts, repository access, and a walk through the deployment path from commit to production. By day three the application runs on the engineer's own machine or in a development environment. The first pull request lands in the first week and is deliberately small — a real change, not a setup commit.

By day 15 we expect three things to be true: the engineer has merged to production at least twice, can take a ticket off your backlog without asking which repository it belongs in, and has written down at least one thing about your system that nobody had documented.

Who pays for the ramp: the first five business days are not invoiced. From day six the engineer is invoiced in full, ramped or not. Past that week, a stalled ramp is usually a missing access request on your side, and the clock is the fastest way to surface it.

On-call is separate. An engineer joins your rota no earlier than 30 days in, and only if you run one.

The working agreement

A two-page document signed alongside the SOW. It fixes five things before anyone writes code.

• Hours and overlap. Engineers on US East accounts work 11:00–20:00 Kraków time, which is five hours against a 09:00–17:00 Eastern day, and six across the three weeks in March and the week in late October when the two sides change clocks on different dates. More is bought an hour at a time by moving the start later: a 12:00–21:00 day is six hours, and seven in those same weeks. US West accounts get 2–4 hours, EU clients the full working day.
• Rituals. Your standup, your sprint cadence, your tools, your board. We do not run a parallel process alongside yours. A written standup note goes into your channel before the overlap opens, so it has been read by the time anyone joins a call.
• Definition of done. Ours unless you have your own: merged to main, behind a flag if incomplete, tests covering the change, deployed to staging, documentation updated in the same pull request, ticket closed by the reviewer rather than the author.
• Code review. Your engineers review our pull requests; we do not approve our own work into your main branch. If you have no reviewer with capacity, our team lead reviews instead and your counterpart gets a list every Friday of what merged without a review from your side.
• Escalation. Engineer, then our team lead the same day, then the account owner within one business day. Anything still open after five business days goes to a written review with both sides and a dated action against each item.

Reporting

Weekly, every Friday by 17:00 CET, in writing, about one screen long: what merged, what slipped and why, days billed per engineer against the SOW, every blocker waiting on your side with an owner and a date, and what is committed for the coming week. Written by the team lead on your account, not generated.

Monthly you get an invoice with a per-engineer day breakdown, a 30-minute call with that team lead, a risk list carried forward rather than rewritten, and an absence and turnover log for the engineers on your account — including anyone who has given notice.

No velocity chart and no burndown; neither survives contact with a team that is also doing production support. Any number you want that is not on that list goes into the working agreement and you get it.

When an engineer leaves

We do not promise it will not happen to you. Annual attrition here is 11% and median tenure is 2.8 years, which across 21 engineers is roughly two people a year. On a four-engineer team running 14 months, the chance that one of those two is on your account is real rather than theoretical. Anyone who tells you their turnover is zero is either very new or not counting.

Four things trigger a replacement, and nothing else does: the engineer resigns; long-term absence such as illness or parental leave; we take them off the account because the work is not good enough; or you ask for them to be replaced, in writing, without giving a reason. Your commercial situation elsewhere in our client list is never one of them.

Within five business days of notice you get candidates for the seat, under the same rules as the first hire — CVs, an interview with your team, your veto. Candidates, not a start date: there is no bench here, so the names come from engineers rolling off engagements that are finishing and from people already through our three-stage hiring process. Both are visible about a month out, because our engineers serve 30 days' notice to us and clients serve 30 days on an exit. Where neither source holds someone who fits your stack, you get that sentence rather than a name, and the seat takes longer than 30 days to fill.

The replacement overlaps the leaver for three weeks. In those three weeks the leaver pairs with the replacement on the replacement's first tickets, hands over the services they own one at a time, transfers on-call, and brings the runbooks up to date. The handover arrives as pull requests in your repositories, not a document on our side you never see.

What it costs you: one seat. You are invoiced for one engineer across those three weeks, not two, and the replacement's first five days are unbilled exactly as at intake. There is no replacement fee and no re-ramp charge. We absorb the overlap because we are the reason it is needed.

When the three weeks are not possible — a dismissal for cause, a sudden illness — you hear it the day we know, in writing. The seat is then covered by the team lead on your account while the replacement ramps, the leaver's unserved notice is not invoiced, and the handover falls back to whatever is in the runbooks. That version is worse than the planned one, and we are not going to describe it as anything else.

Exit

Thirty days' notice, either side, in writing, no reason required. Notice can be served during the three-month minimum but does not shorten it. Ending on a planned date is normal here: one of our case studies is an eight-month handover whose exit date was set in week one.

What you hold on the last day: a runbook for every service we owned, architecture decision records written at the time each structural choice was made, an on-call handover naming who owns which alert and dashboard, and a credential list with a revocation time against every line.

Everything we write lives in your repositories, in Markdown, updated in the same pull request as the code it describes. We keep no wiki on our side, so there is nothing to export at the end.

No lock-in: infrastructure runs in your AWS accounts, Terraform lives in your repositories, no Shyk Systems licence is embedded anywhere, and there is no in-house framework of ours in your stack. The final two weeks are walk-throughs, one per service, with whoever is taking it over.

Security and data handling

A mutual NDA is signed before the first CV is sent, and it covers your identity — which is why no client is named anywhere on this site. Contracting is through an EU entity, and a GDPR data processing agreement is annexed to the MSA: you are the controller, we are the processor, with a named sub-processor list you are notified about before anything on it changes.

• Devices. Company-owned laptops only, full-disk encrypted, centrally managed, screen lock at five minutes. No client code on personal machines. No local copies of production data — engineers work against anonymised or synthetic data, or inside your environment under your access.
• Access. Named individual accounts through your identity provider, never a shared login. Least privilege by default. Production access is granted per person, by you, and can be withdrawn by you without going through us.
• Background checks. A criminal record check and two employment references, completed before an offer is made. Every engineer, including team leads, no exceptions for referrals.
• Offboarding. Access is revoked on the engineer's last day and your counterpart gets written confirmation within two business days listing every system and the time it was removed. Laptops are wiped and re-imaged before reissue.
• Incidents. If we become aware of a suspected breach touching your data, your named counterpart hears within 24 hours — before we know the cause, not after.

There is no certification badge on this page. We hold none, and the practice above is what you would be buying either way.

Commercial terms

An MSA with the EU entity, plus one SOW per engagement naming the engineers, the rate per engineer, the start date, and the notice terms. Minimum engagement is three months, because ramp-up alone runs 10–15 business days and below three months you are paying mostly for the ramp. After that, 30 days' notice on either side.

The indicative rate is from €11,500/month for a two-engineer team. It moves with seniority and technology; what the SOW carries is a per-engineer rate fixed for the length of that SOW, not a band, and it stays indicative until the scope note exists.

Invoicing is monthly in arrears in EUR, against the same per-engineer day breakdown as the monthly report, payable within 30 days. No setup fee, no recruitment fee, no fee to replace an engineer, no charge for the three-week overlap. Travel to your office is invoiced at cost and only when you have asked for it in advance.