~/pankaz ❯ whoami
Pankaz Jha
QA Automation Engineer
72 bpm · uptime: since med school
From the operating room to the CI pipeline.
I build test automation from scratch — UI, API, mobile, performance — and wire it into CI/CD so quality is continuous, not a last-minute scramble.
pankazportfolio
- role
- QA Automation Engineer
- focus
- UI · API · mobile · performance
- stack
- Python · Pytest · Playwright · Appium
- ci_cd
- GitHub Actions · Azure DevOps · Docker
- coverage
- 1,500+ E2E tests · 70% of test base automated
- location
- Kyiv, UA · remote / hybrid
- langs
- Ukrainian (native) · English (B2)
- uptime
- since med school
About
For ten years I was an anesthesiologist — a job built on protocols, relentless monitoring, and catching problems before they ever reach the patient. I bring that exact discipline to QA automation.
Today I build test automation from the ground up — twice as the sole automation engineer on the team. Most recently: 1,500+ Playwright tests for a US construction-management SaaS, Android automation with Appium, JMeter load testing that drove autoscaling, and CI pipelines with diff-based test selection and parallel shards. Before that, I cut a 3-day manual regression down to a 5-hour automated run.
I also build and ship my own products — an Android app, web apps, self-hosted automation pipelines — increasingly with AI-assisted development. Shipping software myself makes me a sharper tester: I know what breaks, because I've built it.
Skills
Automation-first. Development skills are second-tier proof.
Test automation
- Python
- Pytest
- Playwright
- Allure
- Page Object & Screen Object
API testing
- Postman
- Swagger
- Requests
Performance
- JMeter
- Prometheus
- Grafana
Mobile
- Appium + Python (Android)
- Screen Object
- emulators & real devices
Databases
- PostgreSQL
- MsSQL
- SQLAlchemy / SQLModel
- queries · data validation
CI/CD & infra
- GitHub Actions
- Azure DevOps
- Docker
- Docker Compose
- AWS
- Linux / bash
- Git
QA process
- Test design
- Test documentation
- Qase
- Jira
- Agile / Scrum
Also builds
- Android
- Web (Django / Next.js)
- Claude Code skills & AI-assisted development
Familiar with
- k6
- Selenium
- JavaScript
Spoken
- Ukrainian (native)
- English (B2)
Experience
QA Automation Engineer · LNOKS
Client project: EZBuild — multi-tenant construction-management SaaS for US general contractors (web + Android). Sole automation engineer in a 6-person team.
- Built the E2E UI framework nearly from scratch (Python, pytest, Playwright, POM): 1,581 tests and 257 page objects across 7 functional areas; 70% of the ~2,000-case test base automated.
- CI on GitHub Actions: diff-based test selection, parallel shards on isolated tenants, merged reporting; a 137-test smoke suite gates every stage deploy.
- Android automation: ~100 tests with Appium (Screen Object) on emulators and real devices.
- JMeter load tests (9 scenarios, up to 1,000 users) with Prometheus + Grafana, weekly on AWS — exposed bottlenecks that led to autoscaling and DB optimizations.
- Built Claude Code skills for Qase test cases and Jira bug reports — case authoring went from ~10 a day to ~100 an hour; adopted company-wide.
- Verified data in PostgreSQL (SQLAlchemy); worked directly with the US client in English — demos, grooming, weekly syncs.
Quality Control Engineer (QA Automation) · Infopulse
Internal time-tracking system used company-wide (~2,000 employees). Sole automation engineer in a 7-person team.
- Started in manual QA; moved to automation within six months and built the framework from scratch (Python, pytest, Playwright, Page Object).
- ~500 UI and API tests covering 80%+ of critical functionality — regression cut from 3 days to a ~5-hour automated run.
- MS SQL validation after UI actions; DB-level test preconditions and cleanup.
- Azure DevOps pipelines run the suite on every commit.
Anesthesiologist · Kyiv clinical hospitals
Alexander Clinical Hospital, International Neurosurgery Center.
- A decade of precision-critical decisions, protocol-driven work, and high-stakes responsibility under pressure — the foundation of how I approach quality, risk, and verification today.
Education
- Medicine — Bogomolets National Medical University, Kyiv
- Residency — Shupyk National Medical Academy of Postgraduate Education
- Certificate: Fundamentals of Software Testing — QATestLab
Projects
I build real systems end-to-end — so I understand what I test — and I automate everything: scraping, pipelines, CI, infra.
Mirrolit — bilingual e-book reader (Android)
An Android reader that shows the original text and a live translation side by side — EPUB, FB2 and MOBI, on-device translation with ML Kit and offline models. In closed testing on Google Play.
Built with Claude Code and covered by unit and instrumented tests — I test what I build.
- Kotlin
- Jetpack Compose
- ML Kit
- Room
- Claude Code
LabLib — Android app (AI-assisted)
A native Android app that interprets lab results, built largely through agentic AI development (Claude Code) and shipped end-to-end from product planning to deployment.
Proves I can build and reason about the same mobile apps I write automation for.
- Android
- Kotlin
- Claude Code
News pipeline → Telegram
v1: a Python bot — Playwright scraping, RabbitMQ queues, OpenAI API for processing. v2 (live): a self-hosted n8n pipeline on Hetzner — scheduled scraping, translation and filtering, hash-based de-duplication in PostgreSQL, auto-publishing to Telegram.
Rebuilt the system when v1 stopped fitting — owning the whole stack, not just the tests on top of it.
- Python
- Playwright
- RabbitMQ
- OpenAI API
- n8n
- PostgreSQL
- Hetzner
protocols.pankaz.dev
A clean, structured reference site of translated anesthesiology clinical guidelines (e.g. the DAS 2025 difficult-airway guidelines) — recreated as styled HTML with algorithm diagrams.
Brings medical domain knowledge and engineering together, built fast with AI-assisted development.
- Next.js
- Claude Code
“I use agentic AI tooling (Claude Code) to ship real products — two Android apps, a clinical protocols site — and to speed up QA itself: my Claude Code skills for test cases and bug reports took case authoring from ~10 a day to ~100 an hour and are now used company-wide. AI-generated output gets the same treatment as any other code: review, test, verify.”
Contact
Kyiv, Ukraine — open to remote and hybrid.
Open to remote and hybrid roles. The fastest ways to reach me: