Skip to content
View MilJav11's full-sized avatar
1 follower · 0 following

Achievements

Achievement: Pull Sharkx2Achievement: YOLO

Achievements

Achievement: Pull Sharkx2Achievement: YOLO

Block or report MilJav11

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
MilJav11/README.md

Miloš Javorský

QA Automation & AI Engineering

I come from QA automation and now focus on testing and securing LLM/AI systems: evaluation, red-teaming, RAG safety and self-healing E2E tests. I like test suites that run offline, gate merges in CI, and are honest about what they don't cover yet.

Featured projects

  • ai-qa-factory: local, human-in-the-loop workspace that generates Playwright tests, diagnoses failures and proposes locator repairs, which are applied only after explicit approval. Python, Streamlit, Playwright/TypeScript, OpenRouter; CI on Python 3.10–3.12.
  • rag-security-sandbox: RAG app and pytest fuzzer for prompt injection, data poisoning and prompt leaking, with regex + LLM-as-a-judge guardrails. FastAPI, ChromaDB, Groq. Offline test suite in CI; known security gaps are documented as xfail(strict=True) tests.
  • llm-api-benchmark: pytest framework benchmarking local LLMs (via Ollama) on latency, instruction following, LLM-as-a-judge, red-team prompts and RAG hallucinations. Test data in JSON, HTML reports, offline suite in CI.
  • playwright-python-e2e-framework: BDD E2E suite (pytest-bdd, Page Object Model) with a credential-free demo suite gating CI and a manual live-email suite. Python, Playwright, GitHub Actions.
  • triglav-harness: experimental public preview of a local Python coding controller.
  • repo-guard: fast local git guardrail against accidental secret leaks in AI-generated code. Python.
  • prompt-refinery: Architect → Referee pipeline that turns raw tasks into contract-validated prompts for AI IDE agents, with an MCP server, Streamlit GUI and 258 offline tests. Python.

How I work

I build with several AI coding agents – Claude Code as the lead, plus Codex and Mistral (Vibe CLI / OpenCode), in Zed and Antigravity. A human stays in the loop:

  • agents work on branches, never directly on main
  • git hooks block pushes and changes to main
  • every change goes through a reviewed pull request with green CI

Tech stack

Python Pytest Playwright TypeScript GitHub Actions FastAPI Streamlit Ollama

Also: pytest-bdd, ChromaDB, LLM APIs (OpenRouter, Groq).

Contact

LinkedIn

Pinned Loading

  1. triglav-harness triglav-harness Public

    TRIGLAV — Three Heads. One Verdict. Experimental public preview of a local Python coding controller.

    Python

  2. rag-security-sandbox rag-security-sandbox Public

    DevSecOps playground for testing RAG application security — simulates Prompt Injection, Data Poisoning & Prompt Leaking attacks with automated Pytest fuzzing and real-time SOC dashboard.

    Python 1

  3. ai-qa-factory ai-qa-factory Public

    Human-in-the-loop Playwright E2E test generation and safe self-healing with Streamlit, OpenRouter, and DeepSeek.

    Python

  4. llm-api-benchmark llm-api-benchmark Public

    Automated 4-layer benchmarking framework for local LLMs (Llama 3.2, Phi-3, Qwen) via Ollama and Pytest — testing latency, instruction-following, security guardrails, and RAG hallucination prevention.

    Python

  5. playwright-python-e2e-framework playwright-python-e2e-framework Public

    Professional E2E test suite for Seznam.cz email workflow using Playwright, Python, BDD (pytest-bdd), and Page Object Model with GitHub Actions CI/CD integration.

    Python

  6. repo-guard repo-guard Public

    Fast local git guardrail against accidental secret leaks in AI-generated code.

    Python