NOTE 01 / LIVE IN PRODUCTION✷ side project
Pilot
A project and task management SaaS built end to end with my own agentic toolkit: Rails 8 API, React + TypeScript SPA, Kamal 2 deploys, tag-only releases and one-click rollback.
HI! I'M OLEH, NICE TO MEET YOU.
I ask
“why?”
before I write code.
Ruby on Rails engineer with 5+ years across EdTech, UK healthcare and fintech. These days I build AI product features — and the agentic workflow that ships them.
✳ Poland (CET) · Ruby on Rails since 2019 · Remote · B2B
things I keep coming back to:make it work.
prove it works.
prove it works.
currently deep intoRails ↗ agents ↗ MCP
01 / THE WALL OF IDEAS
There's always
something cooking.
Some of these run in production. Others are the tools I built to ship them. All of them taught me something worth keeping.
connected by evidence ✳Follow a thread:
NOTE 02 / THE HOMELAB✓ infrastructure
Self-hosted infrastructure
One server, many isolated VMs: dev environments, a self-hosted GitHub Actions runner and workspaces where coding agents run unattended — reachable over SSH and RDP.
NOTE 03 / HOW I SHIP✳ a living toolkit
Agentic SDLC
My AI-driven delivery toolkit: Claude Code plugins, Archon factory workflows and spec-driven delivery — PRD → TRD → ADR → a tested, reviewed PR.
NOTE 04 / AI FEATURES AT SCALE♡ the product side
AI assistant for 3,000+ schools
At ManageBac I built the backend of an AI assistant chat, one-click prompt chips and LLM token-usage analytics — and shaped MCP tools around the chart queries users already trust.
✳ These are personal stories and purpose-built illustrations. Private code, employer systems and customer information stay private.
02 / ON THE WORKBENCH
Try the rough ideas.
Three tiny experiments you can poke at. No polished magic tricks here — just the thinking made tangible.
03 / THE PERSON BEHIND THE TABS
A few notes about me.
PERSONAL NOTE / FROM COMPUTER ENGINEERING TO AGENTS
I studied computer engineering (M.Sc., Khmelnytskyi National University) and came to Ruby on Rails in 2019. Since then I've worked on systems where mistakes are expensive: a national UK NHS dental triage service that routed 100K+ referrals, an IoT platform at FluentPet, and a crypto exchange.
At ManageBac — a learning platform used by 3,000+ IB schools — I built the backend of the AI assistant chat, prompt chips and token-usage analytics in a 10+-year Rails monolith. Along the way: pentest findings fixed across 3 vulnerability classes, test coverage up 40%, p90 query latency down 15%.
Now I build Pilot end to end with my own agentic SDLC toolkit. My favorite part hasn't changed: ask why first, reuse what already works, and only call it done when the evidence says so.
— Oleh ↗
04 / RECEIPTS
Certificates I can back up.
- ANTHROPIC EDUCATION · OCT 2026
Model Context Protocol: Advanced Topics
Verify credential ↗ - ANTHROPIC EDUCATION · OCT 2026
Introduction to Model Context Protocol
Verify credential ↗ - SULICOM AI ACADEMY · OCT 2026
Workshop — Agentic Engineering Workflows
ID WS-KB8M2SBQ59Verify credential ↗ - SULICOM AI ACADEMY · SEP 2026
Workshop — Harness Engineering
ID WS-ZDEFDNMYLPVerify credential ↗ - ENGLISH4IT · AUG 2026
English for Tech
View credential ↗ - RUBYGARAGE · PIVORAK · 2020
Ruby on Rails — complex course & summer course with practice
05 / OPEN ANOTHER CONVERSATION
Got something
to ship?
A Rails codebase that needs care, an AI feature that has to earn its cost, or a team adopting coding agents — I'm open to remote B2B work and good conversations.