Yaniv
Metuku
>
2 PyPI packages shipped. 53 automated tests across 4 layers. AI-powered failure analysis built in.
System Architecture & Mindset
I approach quality assurance as a strategic business function, not just a checkbox. With expertise spanning API testing, CI/CD automation, and AI-powered failure analysis (using Groq LLM), I build testing ecosystems that catch issues before they reach production. I focus on "Shift-Left" methodologies, ensuring every commit transitions software from 'working' to 'bulletproof'.
GitHub Activity
Tech Arsenal
Test Automation
API & Infrastructure
AI Integration
DevOps & Management
QA Methodologies
Featured Automations
Real-world testing architectures built to intercept failures before they reach production.
Financial Integrity Ecosystem
End-to-end test automation framework that mathematically proves data integrity across a financial system (Expense Tracker) — validating that data entered via the UI arrives intact in the database. Not just status 200. Built solo: architecture, design, and execution.
Architecture Highlights
- ▹ 4-Layer coverage: Web (Playwright), API (Flask+Requests), Mobile (Appium/UiAutomator2), Database (MySQL+SQLite)
- ▹ Set Theory validation: new_set - old_set = exact record added — mathematical proof of data completeness
- ▹ AI Failure Analysis: automatic root-cause analysis on every test failure via Groq LLM
- ▹ DDT: Data-Driven Testing with external CSV + JSON — zero hardcoded values
- ▹ CI/CD: GitHub Actions → Allure Reports → GitHub Pages
- ▹ Mobile tested on a real physical device — not an emulator
Live Metrics
Failscope
A published PyPI package (v0.1.1) that auto-analyzes test failures using a dual-agent AI pipeline powered by Groq/Llama. Reads pytest logs, classifies failure type, and generates structured bug reports — all from a single CLI command.
Architecture Highlights
- ▹ Dual-agent pipeline: Agent 1 classifies failure type, Agent 2 generates actionable resolution
- ▹ Published to PyPI v0.1.1 — installable via pip install failscope
- ▹ Smart Log Preprocessing: Cleans and structures pytest logs before LLM analysis
- ▹ Zero data leakage option: supports local Ollama for offline inference
- ▹ Generates structured bug report templates ready to paste into Jira
Live Metrics
FixtureForge
Stop writing 'Test User 1' and 'Lorem Ipsum'. FixtureForge is an Agentic Test Data Harness — deterministic in CI, AI-powered in development. Provider-agnostic: Claude, GPT, Gemini, Groq, Ollama, or no AI at all.
Architecture Highlights
- ▹ Two Modes: seed=42 for fully deterministic CI output, AI-powered context for development — same API, different fidelity
- ▹ Smart Field Routing: only semantic fields (bio, description, review) hit the AI. name/email/phone use Faker for free
- ▹ DataSwarms: generate multiple models in parallel with shared AI cache — 5 models cost ~1.5x, not 5x
- ▹ Permission Gates: safe / sensitive / dangerous classification with human-in-the-loop for PII and security test data
- ▹ Lazy Streaming: generates 1GB+ datasets record-by-record without loading into memory — exports to JSON, CSV, SQL
- ▹ Smart Relationships: automatically links child records (Orders) to parent IDs (Customers) — no manual FK wiring
Live Metrics
Network & Security QA Lab
Hands-on quality engineering across VPN infrastructure, packet analysis, and network-probe APIs — built to validate the real-world behaviour of network and security systems.
NetProbe-QA
Automated quality gate for a REST network-probe API. Validates endpoint correctness, error handling, and data integrity using pytest + Requests. Covers status codes, JSON schema, boundary values, and fault-injection — CI runs on every push via GitHub Actions with full Allure reporting.
Technical Highlights
- ▹Schema validation: every response field type-checked against a Pydantic model
- ▹Fault injection: tests for 4xx/5xx behaviour, malformed payloads, and empty bodies
- ▹Boundary-value analysis on numeric probe parameters
- ▹CI/CD: GitHub Actions runs the full suite on push; Allure report published to GitHub Pages
- ▹Parametrised fixtures — zero duplicated test logic across 20+ cases
Live Metrics
PacketSentry
Automated validation suite for a packet-capture and traffic-analysis service. Tests inspect live-capture correctness, filter logic, protocol parsing, and alert thresholds — verifying that zero packets are dropped or mis-classified under load. Full pytest suite with Docker-isolated capture environment.
Technical Highlights
- ▹Live-capture assertions: confirms packets are captured, not just acknowledged
- ▹Filter-logic tests: BPF expressions validated against known traffic patterns
- ▹Protocol parsing: UDP / TCP / ICMP frames parsed and field-level verified
- ▹Threshold tests: alert fires at configured packet-rate boundary — not before, not after
- ▹Docker-isolated environment: repeatable captures with zero host-traffic noise
Live Metrics
VPN-QE-Lab
Self-contained IPSec VPN quality gate: spins up two strongSwan IKEv2 gateways in Docker, establishes a full site-to-site tunnel, then runs 9 automated tests covering IKE Phase 1 & 2, bidirectional routing, traffic encryption, and NAT Traversal (ESP-in-UDP). Mirrors a real-world SD-WAN deployment scenario.
Technical Highlights
- ▹IKE Phase 1 & 2: asserts ESTABLISHED + INSTALLED SA state via ipsec statusall
- ▹Bidirectional routing: ping 100.64.1.1 ↔ 100.64.2.1 through encrypted tunnel — 0% packet loss
- ▹Encryption proof: tcpdump on WAN confirms UDP/4500 traffic — zero plaintext ICMP
- ▹NAT-T validation: ip xfrm state show asserts espinudp encapsulation in kernel
- ▹forceencaps=yes simulates SD-WAN behind carrier NAT (ESP protocol 50 blocked)
- ▹CI/CD: full compose up → await ESTABLISHED → run 9 tests → compose down on every push
Live Metrics
Experience & Education
A track record built on operational discipline, technical training, and independent engineering.
Robotics Instructor (Volunteer)
Hamapilim School, Lod
Designed and delivered robotics and programming lessons for 2nd-grade students as part of a STEM enrichment programme.
- Built structured lesson plans and hands-on exercises for groups of 5-7 students.
- Applied the same documentation mindset used in QA: clear steps, expected outcomes, and verification.
- Demonstrated ability to break complex technical topics into verifiable, reproducible steps.
Planner & Logistics Controller(Deputy Team Lead)
IDF — Masha 7200
Managed end-to-end inventory processes in a mission-critical military environment using SAP ERP.
- Owned data integrity for mission-critical inventory — identified and resolved discrepancies in real-time.
- Coordinated with suppliers and production teams under strict operational deadlines.
- Maintained 100% accountability across high-value asset tracking using SAP ERP.
Operations & Logistics Specialist
Various Companies (Manpower)
Executed logistics and inventory operations with high attention to detail and process discipline.
- Demonstrated consistency and reliability across multiple high-throughput environments.
- Applied structured process thinking from industrial management background.
QA Automation & AI Bootcamp
Tech-Career
Intensive 800+ hour program covering Manual Testing, Automation, and AI-assisted QA.
- Built a comprehensive E2E test automation framework using Python (OOP), Pytest, and Playwright.
- Designed and executed REST API tests with Postman and integrated database validation via SQL.
- Managed test artifacts, bug lifecycle, and documentation in an Agile environment (Jira, Xray).
P.E. Industrial Management
The Joint College
Practical Engineering degree specializing in Operations Management and Information Systems.
- Specialization in Operations Management and Information Systems.
- Strong foundation in process optimization and data-driven decision-making.
Let's Connect
Open to QA Automation roles and freelance testing projects. Reach out directly or check out my live work below.