The menu
One contractor, the full quality stack. Bring me in for a single piece or the whole thing. Each of these is a problem I have already solved in production.
A flaky suite everyone has learned to ignore is worse than no suite. I build automation that holds across the whole stack - Cypress, Playwright, Selenium, pytest, JUnit, Jest - with page objects, stable selectors, and seeded state. Green means green.
Shipping features on GPT or Claude and hoping they behave? I build the evaluation layer that proves they do: output scoring against your ground truth, hallucination and regression gates, prompt-injection and jailbreak red-teaming, and RAG evaluation. Built on the Anthropic and OpenAI APIs with Promptfoo, Braintrust, and DeepEval.
Deploys that make everyone nervous usually mean tests do not run automatically. I wire the pipeline - GitHub Actions, GitLab CI, Jenkins, Docker - so every pull request is tested and a red gate blocks the bad ones. Boring releases by design.
When the suite is too slow or the runners are flaky, testing gets skipped under pressure. I move it to the cloud - containerized, parallel, self-hosted or managed runners on AWS or GCP - so a full nightly finishes under two hours.
No process and bugs keep reaching production. I set the strategy: risk-based test plans, release-readiness criteria, sign-off, and post-release monitoring. The gate is mine to hold.
Automation cannot cover everything. I run the exploratory passes and UAT that catch what scripts miss, before your users do.
Inbox open
Mail is the fastest way to reach me. Tell me what you are shipping and where it hurts, and I will tell you how I would test it.
Email tykhonkozachenko@gmail.comT.