// team
There is one of us. You should know that before you depend on it.
A team page is usually there to tell you the thing is safe to rely on. For a platform built by one person the honest version of that starts with the size of the team, not with a CV — so the cost is first, and everything after it is either something that lowers the cost or something you can check without asking me.
// what that costs you
No hosted offering, no support contract, and a bus factor of one
There is no signup, no service level and nobody else to escalate to. If this stops being maintained, it stops. Nothing below changes that, and anyone telling you a one-person project has no key-person risk is selling something.
// and to be exact about the count
The two sessions on the history page are not a second person
They are two AI sessions working on separate parts of the tree, keeping a shared record and re-measuring each other's claims. That is a working method and it is described where it belongs, on the history page. It is not a team, and counting it as one would be the first dishonest thing on this site.
// what lowers the cost
You are not locked in, and that is structural rather than a promise
What ships is not one repository you would have to inherit whole. Every package is its own repository with its own remote — the list is at github.com/coolms, and it is the list rather than a number quoted here, because a number quoted here would be wrong by the time it mattered. Forking the parts you depend on needs no permission and no cooperation from me. That is a property of how the thing is packaged, not an undertaking I am giving you.
The reasoning is written down rather than remembered
// checkable without asking meA record for every load-bearing decision
Every load-bearing choice has a numbered record saying what was decided, what it cost, and what was rejected — including the ones later revised, which keep their original text and carry the revision beside it.
Every task closes with what was measured
Every task closes with what was measured. The wrong measurements are in there too, corrected in place with the correction dated — one of them written the day this page was built.
More test files than a coverage percentage would tell you
Not a coverage percentage, which can be met without testing anything. Two counts you can run yourself in the tree.
Where to look, if you would rather check than take my word. Counted as COMMITTED on 2026-09-10 — this tree and, where a row says so, the package repositories too, because code moving from one to the other does not make the platform smaller.
what where how much
// the person
The same person wrote the 2006 CMS, the 2011 rewrite, and this one
The template class in the 0.x archive is stamped "Copyright (c) 2003-present Dmitry Popov", which is the name on the commits in this tree. That is not a seniority claim — it is why the same three ideas keep coming back, and why the second attempt's failure is described on this site in more detail than its successes.
// the CV, and what it is evidence of
The first CMS here was written in the newsroom the history page starts in
The 2000–2006 line below is a news agency and a national newspaper — the newsroom that made this a filesystem instead of a role matrix, and where the Perl text processors the history page says are gone were written. The 2006–2016 line is the company CoolMS.CMS was built to run. That is the part worth checking rather than taking: the platform's story and the person's are the same story, and each can be read against the other.
Twenty-six years, from the newsroom to this. Roles and dates as they stand on the CV of 2026-08-26; the platform is the last line rather than the first.
years role where
Not the whole CV — the parts a reader of this page can use
// the three that bear on this platform specificallyRan a department, then a practice
Head of PHP at Innowise, where a Go practice started inside the PHP department and was spun out as its own department at around twenty engineers. The claim on this site is not that one person is enough; it is that this one has worked at the other scale too.
Architecture as the job, not a side effect
Solution Architecture training on IASA and TOGAF practice, and a Solution Architecture Document — C4 views, target state, phased plan — for a payment platform's monolith-to-microservices move. The boundary rule on this platform is that habit, enforced.
Delivered inside somebody else's large codebase
A scheduling capability built from scratch into a live product — three domain modules, 450+ classes, 47 endpoints — designed so any existing entity became schedulable through a port, with no changes to the calendar module. Greenfield is easier than that.
// and what it is not evidence of
A track record is not a guarantee, and one person is still one person
Everything above says this has been done before. None of it changes the first thing on this page. Read it as a reason to expect the work to hold up — not as a reason to expect it to continue.
// where to find it
Source-available, self-hosted,
and answerable in one place.
Questions go to GitHub discussions, where the answer is public and stays there. There is no support address, because an address implies someone is watching it.