Pavel Anhur

Professional Summary

QA Automation Engineer with 4 years of experience designing, building, and maintaining test frameworks for complex web applications. At AlgoSec, I own the automated test strategy for the security and firewall management domain — a product where a missed regression has real security consequences. My main stack is TypeScript and Playwright, complemented by Java and Python. I focus on test suites that are fast, stable, and readable enough that developers run them locally before pushing. My portfolio includes a public Playwright + TypeScript framework with 250+ tests, and a five-week LLM test harness covering judge calibration, RAG metrics — including a real retriever and streaming. Both have published Allure reports. I also built an MCP-based AI layer with four agents — requirements analysis, failure triage, adversarial review, and test generation with a heal loop.

Core Skills

Languages: TypeScript, Java, Kotlin, Python, SQL (MySQL, PostgreSQL)
Test Automation: Playwright, Selenium, REST Assured, JUnit, Pytest
AI in QA Workflows: AI-assisted test generation, LLM log analysis, MCP multi-agent workflows, prompt engineering, local models (llama.cpp, Ollama)
LLM Testing: Property-based assertions, pass-rate thresholds, adversarial inputs, LLM-as-a-judge, RAG metrics (faithfulness, relevance, context precision, recall), vector retrieval (Qdrant, embeddings), streaming tests
Framework Design: Page Object Model, custom fixtures, step decorators, data-driven tests, test tagging, visual regression
CI/CD: GitHub Actions, Jenkins, GitLab CI
Reporting: Allure, Playwright HTML, JUnit XML
API Testing: REST, OpenAPI/Swagger, Postman, contract testing basics
Infrastructure: Docker, Linux (incl. SSH on production), AWS (basic)
Version Control: Git, GitHub, Bitbucket, Jira
Build Tools: npm, Maven, Gradle

Experience

Automation Engineer · AlgoSec · Prevasio

Jan 2023 – Aug 2024

Prevasio is AlgoSec's cloud security product — a CSPM solution covering container security, vulnerability assessment, and automated cloud network security. I joined as the first and only automation engineer, alongside one architect, three developers, and a small product team.

Combined manual and automated testing — ran exploratory manual sessions alongside the automated suite to catch issues automation didn't cover.
Built the automation practice from zero for a greenfield CSPM product — chose the stack, defined the test strategy, and wrote the first test before the UI had stable locators.
Automated 300+ tests across UI and backend for cloud security scanning, vulnerability assessment, and container posture workflows.
Defined the test strategy for a product with no existing test process — identified the highest-risk workflows, wrote the coverage plan, and set the standard for what "tested" means on the team.
Covered multi-cloud workflows end-to-end across AWS, Azure, and GCP — credentials, scanning, findings, remediation — catching cross-provider regressions that single-cloud testing would have missed.
Built visual regression coverage (screenshot comparison) into the suite.

Automation Engineer · AlgoSec · Devices

Jan 2022 – Dec 2022

AlgoSec's core security policy management product for enterprise firewalls. I was on the Devices team, responsible for the firewall policy change workflow, until moving to the Prevasio product in January 2023.

Owned the automated test suite for the firewall policy change workflow — 400+ UI and API tests covering policy creation, validation, deployment, and rollback.
Authored test designs and reviews for the security-policy domain, collaborating with developers to catch spec-level ambiguities before code was written.
Reduced suite runtime and flakiness through targeted waits, and elimination of cross-test contamination.

Automation Test Engineer · EPAM Systems

May 2021 – Dec 2021

Automated test cases for web services, mobile, and web applications using Java and Selenium.
Worked alongside developers to keep automation aligned with application changes; debugged failures and reported on suite health.
Produced detailed test result reports for stakeholders across the client and delivery teams.

Projects

Playwright + TypeScript Test Framework

github.com/PavelAnhur/playwright-typescript-framework

A public reference framework for end-to-end testing. 250+ tests across UI and API, organized by spec, with custom fixtures, typed page objects, a custom @Step decorator for Allure reporting, and CI that publishes a live Allure report to GitHub Pages on every push.

An AI layer built on top: an MCP server and four agents — a requirements analyst, a failure analyst, an adversarial judge that reviews the failure analyst's output, and a spec generator that produces tests from requirements and refines them in a heal loop until they pass.

Live Allure report: pavelanhur.github.io/playwright-typescript-framework

LLM Test Harness

github.com/PavelAnhur/llm-test-harness

A test harness for non-deterministic LLM outputs, built over five weeks and documented week-by-week. Property-based assertions replace exact-match checks; pass-rate thresholds gate multi-run suites; adversarial input tests cover prompt injection, role confusion, and payload splitting. Includes a rubric-based LLM-as-a-judge calibrated against 25 hand-labeled examples, and a real RAG retriever — Qdrant vector database, nomic-embed-text embeddings via Ollama, and a deterministic chunker — with faithfulness, relevance, context precision, and context recall metrics. Streaming tests cover token delivery, time-to-first-token, and mid-stream disconnect. Every finding and calibration round is written up, including the ones where the judge disagreed with the human label.

Live Allure report: pavelanhur.github.io/llm-test-harness

Education

Belarusian National Technical University — Minsk, Belarus 2002 – 2008

Languages

Russian (native) · English (professional B2+) · Belarusian