FAQ
The things people ask most — mostly about money and timing. If your question is not here, write to me and I will answer personally.
Frequent questions
Do you work with legacy, or only greenfield?
Both. Breaking a monolith into services and migrating to Kubernetes is one of the most common jobs I take. We start with an audit, so that what already works does not break.
What budget can I come with?
Working out the problem and estimating it costs nothing and commits you to nothing. After that the bottom rung is diagnostics at 3 000–5 000 ₽, an audit from 20 000 ₽, infrastructure end to end from 50 000 ₽, and keeping an eye on a running system from 25 000 ₽ a month. If your problem does not fit even that, I will say so — sometimes it is cheaper to solve it another way.
How is the cost worked out?
In five steps. A 40-minute brief, then a from–to range by email within a day or two — both free. For larger work an audit or design phase sits between them, and its price is credited against the project. Then a fixed quote: the base package plus add-ons and multipliers, all of them listed openly on the pricing page. The quote is fixed before the work starts; if the scope changes we re-agree it in advance rather than after the fact. Payment is 30% to start, 40% midway, 30% on acceptance.
Who pays for the cloud and the licences?
You do, directly from your own account. They are never part of the quote — not the server bills, not the domains, not the paid APIs. That keeps the true cost of ownership visible, and you pay no markup on an invoice I would merely be passing on.
Can I start small and grow?
Yes, the packages are a ladder. “Mini” is one server without Kubernetes, in 3–5 days; once there are more services it moves up to “Start” with credit for what is already done. Nothing is set up from scratch again: the orchestrator goes on top.
Whose repository and whose infrastructure?
Yours. Code, infrastructure as code and documentation live inside your perimeter from day one. There is no lock-in to me.
Do you work alone or with a team?
It depends on the size. A small service I carry end to end myself. For bigger work I have a team, and I bring in extra specialists when a project needs them — I stay the architect and tech lead. There is a third option too: I join your team as a DevOps engineer or a developer, for a project or for a fixed term.
Does AI have to go out to an external API?
No. I can run models inside your own perimeter — for sensitive data that is the standard option.
What does it cost to keep AI running?
Budget 30–40% of the build price a year. That is not a markup but the market norm: models get updated, prompts and the knowledge base need work, and without it the quality of the answers drifts. I say it up front so the number is not a surprise six months in.
How long does the work take?
It depends on the package, and the timing is stated for each one. Infrastructure runs from 3–5 days for “Mini” to 6–8 weeks for “Production”. A Telegram bot takes 3–5 days, an AI assistant over your documents 2–3 weeks. The timing goes into the quote and is fixed along with the scope.
What if the load grows tenfold?
That is settled at the architecture stage: horizontal scaling, queues, caching. Load testing happens before launch, not after it.
What is left when you hand over?
A running system, the repository, infrastructure in code, monitoring with alerts, documentation, and an hour or two of handover to your team. Plus three months of warranty: if something on my side breaks, I fix it at no charge.
Didn't find your question? Ask it and I will answer personally.