Product foundations and the first engineering team
Responsibility Backend, frontend, realtime workflows, deployment, and product foundations. Built and led an early engineering team of five.
Nikolay “Nik” Shurkalin / Warsaw, Poland
Your SaaS should not needthe founder to keep production together.
I help small SaaS teams find the operational risks that matter, improve one critical workflow, and build a safer way to ship. The work stays connected to evidence, code, release decisions, and verified production behavior.
For founder-led SaaS companies with a live product where releases, recurring incidents, or production cleanup still require founder attention.
- Experience
- 13+ years building and operating software
- Built & led
- An early engineering team of five
- Production scope
- 20+ APIs maintained and extended
- Based
- Warsaw, Poland
01 / EXPERIENCE
SELECTED EXPERIENCE
Experience tied to responsibility.
A concise public record of responsibility, scope, and evidence. Client-confidential details stay private.
APIs, integrations, and high-load backend systems
Evidence Maintained and extended 20+ production APIs and delivered 10+ new APIs in one global product environment.
From architecture decisions to runtime behavior
Scope Architecture, backend, cloud, CI/CD, reliability, AI-assisted workflows, and connected-product boundaries.
02 / SPRINT
ONE PRODUCT / ONE CRITICAL WORKFLOW
Make production
less founder-dependent.
The 14-Day SaaS Stabilization Sprint turns recurring production uncertainty into evidence-backed priorities, one or two bounded improvements, regression protection, and a clear 30-60 day sequence.
The first stepDescribe what keeps pulling you back into productionI reply with fit, no-fit, and the smallest useful next step.↗14-Day SaaS Stabilization Sprint
Map one critical workflow, rank the top five risks, diagnose the strongest root cause, and implement up to two safe bounded improvements when evidence and approval allow.
DIRECT
You work with me, not a delegated delivery team.
ACCESS
NDA, evidence, and access boundaries are agreed before work begins.
TEAM TIME
A qualification call, focused evidence access, timely reviews, and a final readout.
NEXT STEP
Send one recurring production problem. I reply with fit, no-fit, and the next useful step.
03 / EVIDENCE
ANONYMIZED CLIENT WORK
What changed after the decision.
Real systems, findings, initiatives, and outcomes. Client identities and sensitive implementation details are removed. See all evidence
01 / PRODUCTION AI
Reliability became a release property.
Quality gates, a recoverable recording pipeline, AWS operating controls, and content-free usage telemetry.
Read case 02 / GLOBAL APIsThree streams became one delivery system.
Shared templates, CI/CD, automation, and mentoring replaced duplicated release work and unclear ownership.
Read case 03 / ANALYTICSA fragile platform became changeable.
TypeScript, pipelines, cloud and database discipline, and a maintainable integration boundary.
Read case 04 / CONNECTED PRODUCTA working product received a trust boundary.
An evidence-backed review separated immediate containment from structural remediation without recommending a rewrite.
Read report 05 / DEVICE SCALEPlatform scale stopped limiting device growth.
A rounded historical signal of approximately 10,000 one-star iOS reviews informed a redesign that removed the practical fleet ceiling.
Read case 06 / SERVERLESS PLATFORMOne catalog, built from an empty repository.
A greenfield AWS application core uses Lambda, on-demand DynamoDB, Terraform, and CI/CD, with a separate containerized proxy for large-file delivery.
Read case04 / WRITING
WRITING
Notes from production.
Ideas and operating notes on architecture, reliability, and delegated work. Browse all writing
NOTE.01
How I use AI agents without outsourcing engineering judgment
A practical AI-assisted architecture workflow using Codex, Claude Code, independent reviewers, evidence, and explicit human decision points.
OPEN NOTE.02A practical AI coding workflow with Codex, Claude Code, and reusable skills
How reusable skills, repository guidance, specialized agents, and review gates make AI-assisted engineering repeatable rather than improvisational.
OPEN NOTE.03From demo to production: what usually breaks first
A practical map of the first reliability, workflow, and observability gaps I look for.
OPEN05 / CONTACT
OPEN CHANNEL
Bring one recurring production problem, risky release path, or fragile user or revenue workflow. I will tell you whether it fits the Sprint and what evidence we need next.
No checkout, commitment, or automatic scheduling. LinkedIn and GitHub remain available as professional context and public evidence.