Backend architecture
Where I am strongest. Data models that survive the next feature, APIs that are boring to consume, queues and caching where they belong. No N+1 queries, no per-row loops — bulk and set-based by default.
Open to new projects
Vladyslav Kryklyvenko
Writing production code since 2016 — I started with hand-coded markup, moved into PHP and Laravel, and grew into architecture and tech lead work. Today I design systems, own the hard parts of the backend, and cover the frontend when a project needs it.
01 — Selected work
Mostly service and outsourcing teams, which means a lot of different domains, codebases and stacks. That variety is the point — it is what makes onboarding into an unfamiliar project routine instead of risky.
SaaS that runs Swedish joint-property associations (samfälligheter) — member and share-ratio registries, billing ledgers, facilities and long-term maintenance plans. I own the architecture and build the key features end to end, from the data model to the screens the board actually uses.
Lead conversion platform. Backend work around lead capture, routing and the integrations that keep sales pipelines in sync.
Affiliate and referral marketing platform for ecommerce brands — commission logic, payouts and high-volume tracking data.
Real-time news, commentary and analysis for credit markets. Content-heavy platform where search, indexing and delivery speed decide the product.
Motion analysis and performance insights for athletes. Sensor data turned into metrics coaches can act on.
One-stop shop for businesses, providing integrated technology services across several product lines.
02 — How I work
Ten years of mostly non-product companies taught me one skill above the rest: reading someone else's system quickly, understanding why it is the way it is, and making changes that do not break it.
Where I am strongest. Data models that survive the next feature, APIs that are boring to consume, queues and caching where they belong. No N+1 queries, no per-row loops — bulk and set-based by default.
A feature is not done at the API boundary. I take it through to the interface — Vue, TypeScript, accessible markup — so nothing is left waiting on someone else's sprint.
New domain, undocumented legacy, a stack I have not touched in a year — I map it, ask the questions that matter, and open a useful pull request in the first week.
03 — AI-augmented engineering
Modern delivery speed is not optional any more. I run Claude and Copilot inside my daily loop, and I keep the same review bar I always had. The model drafts; I decide what ships.
Faster exploration. Mapping an unfamiliar codebase, tracing a call path or summarising a legacy module takes hours instead of days.
Boilerplate and migrations. Scaffolding, repetitive refactors and mechanical rewrites get delegated, so my attention goes to design decisions.
Tests and review. A second pass over every diff — edge cases, error paths and the questions a tired reviewer would miss.
AI features in products. LLM integrations, prompt design and the plumbing around them — retries, cost control, guardrails, evaluation.
04 — Stack
Technologies I have worked with across a decade of production projects.
Backend
Data
Frontend
Infrastructure
AI & process
05 — Journey
Ten years in one direction. No career pivots — every layer I learned made the layer above it cheaper to build, which is why I can now hold a whole system in my head instead of just my slice of it.
I started in 2016 turning designs into markup. Nobody calls that architecture, but it is where I learned that structure is a decision: semantics, document order, how a layout behaves when the content is twice as long as the mockup promised. It is still the reason the interfaces I ship are accessible and fast, not merely finished.
Then the centre of gravity moved server-side and stayed there. Data models, business rules, third-party integrations, migrations you cannot roll back once they have run. This is the work I am best at, and it taught me the two lessons I still lead with: most bugs are data-shape problems, most slowness is query-shape problems, and both are decided long before anyone writes a controller.
Vue on the front, Docker and CI/CD underneath — not to collect logos, but to delete handoffs. A feature I can follow from the migration through the API to the component and into a deployment never sits in a queue waiting on someone else's sprint. On a small team that single difference is most of the timeline.
Today the work is mostly decisions: what the system needs to look like in a year, which shortcuts are worth taking, which ones get paid back with interest. I make the architectural calls, build the parts that are hardest to get right, and review the rest. AI tooling runs inside that loop — it raised my baseline speed without touching the bar for what actually ships.
06 — Contact
Tell me what you are building and where it hurts. If I am not the right fit, I will say so quickly.