I build platforms end to end — and keep them alive for years
I'm Sergei Shapkin — a senior web engineer with more than twenty years spent building, scaling and maintaining platforms that real people use every day. I work across the whole stack: domain modelling and database design, application logic in Laravel and Drupal, front-ends in Vue, the containers and Kubernetes clusters everything runs on, and increasingly the AI tooling that automates the work around it.
I'm most useful where a system has to stay fast and reliable under real load for a long time — news networks serving millions of readers a day, custom CMS engines, business automation, e-commerce catalogs. I don't hand a project off and disappear; I tend to stay with the platforms I build, upgrading and hardening them for years.
Principles I build on
Code that bends, not breaks
I lean on SOLID principles so systems stay open to change: small single-responsibility units, dependencies pointed at abstractions, modules you extend without rewriting. It's what lets a platform survive a decade of new requirements.
One source of truth
Don't Repeat Yourself isn't about clever abstractions — it's about one obvious place for every piece of logic and content. Less duplication means fewer bugs and changes that land everywhere at once.
From schema to cluster
Architecture, application, front-end and infrastructure are one continuous system. Knowing all of them lets me make the right trade-off at the right layer, instead of pushing the problem somewhere else.
Maintainability over cleverness
I optimise for the engineer who reads the code in three years — often me. Boring, well-documented code that upgrades cleanly beats a clever one-liner every time.
Good engineering is mostly restraint
Most of the lasting value in a platform comes from decisions about what not to build. Off-the-shelf when it fits, custom only where the domain genuinely demands it, and a hard line against accidental complexity. I'd rather ship something boring that runs for ten years than something clever that needs a rescue in eighteen months.
But pragmatism cuts both ways. On Afisha.rs, Laravel's conventions didn't fit a content-heavy catalog, so I built a Drupal-style entity-and-term engine on top of it — independent linkable fields, compiled entities cached in Redis, a headless API for the mobile app to come. Principles like SOLID and DRY aren't dogma; they're the tools I use to keep that kind of freedom from turning into a mess.
The best platform is one nobody has to think about — it just stays fast, stays up, and bends to whatever the business needs next.
Have a platform to build, rescue or scale?
Tell me about the system you're planning. I reply within one business day with a clear, honest assessment.