Javier CapaQuality Engineering

Senior Quality Software Engineer

Quality is
an engineering
capability.

I turn risk into validation, automation into evidence, and test results into decisions people can explain.

Explore my work GitHub / @javicale

01 / Selected work

Five projects.
One quality mindset.

Reference implementations, learning labs and private product engineering.
Public projects link to their source.

01

Strategy & governance

Quality Engineering
Playbook

A practical structure for deciding what to validate, why it matters, and what evidence a release needs.

Reference material

How does business risk become a defensible validation decision?

The implementation

Risk-based testing, a QA validation framework, evidence templates and operating principles for shared quality ownership.

What to inspect

The path from change and risk analysis to execution evidence, residual risk and a human release decision.

Read the case study

The problem

Test execution alone does not explain whether the important risks were covered or what uncertainty remains.

The decision

Define validation intent and evidence expectations before execution. Keep residual risk visible when communicating release readiness.

The boundary

This is reusable reference material. It does not claim adoption results, client outcomes or a certified governance standard.

Explore the playbook
02

Automation engineering

Playwright Quality
Engineering

A compact automation reference built around maintainability, diagnostic evidence and CI feedback.

Reference implementation

How can automated checks stay understandable and useful when they fail?

The implementation

Playwright + TypeScript, Page Objects, UI and API checks, and execution across Chromium, Firefox and WebKit.

What to inspect

Test strategy, reusable page abstractions, HTML and JUnit reports, and browser evidence retained by GitHub Actions.

Read the case study

The problem

A growing collection of scripts can be difficult to maintain and provide little context when a check turns red.

The decision

Separate UI abstractions from test intent, keep API checks distinct, and preserve traces, screenshots and video for diagnosis.

The boundary

This is a compact, production-minded reference, not a claim of production deployment. Authentication fixtures, accessibility and visual regression are planned extensions.

Explore the framework
03

AI + data + release evidence

Agentic Quality
Engineering Lab

Exploring agent-assisted planning while deterministic controls retain execution and evidence boundaries.

R&D / learning lab

Can AI expand validation capacity without weakening evidence or accountability?

The implementation

MCP context, plan evals, CSV/SQL differential validation, capability mapping and plan-to-execution traceability.

Observed in the V4 experiment

A live plan scored 90/100; all 5 planned scenarios mapped to registered capabilities and passed. An unsupported HIGH scenario forces NO_GO in the negative control.

Read the case study

The finding

A good plan is not proof of executed coverage. An early SQL experiment exposed that gap, leading to explicit scenario-to-capability mapping and a coverage gate.

The evidence

The documented September 9, 2026 V4 live run separates plan quality, execution permission, scenario evidence and the reference release signal.

The boundary

Small synthetic experiments, illustrative thresholds and human release authority. No claim of production ROI, superior AI judgement or long-term Agentic AI expertise.

Read the verified V4 experiment ↗
Explore the research lab
04

Test observability

Quality
Observability Lab

Connecting a failed browser test to the backend operation and evidence that explain it.

COMPLETED / R&D scope closed

The checkout failed. Can we show where, why, and which test it belongs to?

The implementation

Playwright → test_run_id/traceparent → OpenTelemetry → OTLP/HTTP → OpenTelemetry Collector → Tempo/Loki/Prometheus → Grafana.

Verified final state — COMPLETED

A deterministic evidence gate and both CI contracts, deterministic-evidence and otel-stack-contract, verified on main.

Read the case study

The finding

Matching a trace ID is only the start. The gate also verifies parent-child relationships, the log’s span ID, exception evidence and scenario metrics.

The evidence

The completed lab connects browser execution to correlated telemetry through OTLP/HTTP and the OpenTelemetry Collector, with traces in Tempo, logs in Loki, metrics in Prometheus and inspection in Grafana. The deterministic-evidence and otel-stack-contract CI contracts are verified on main.

The boundary

This laboratory is closed within this R&D scope. It is not a production-ready platform. Synthetic validation and verified CI contracts do not establish production readiness or production SLOs.

Inspect the CI contracts on main ↗
Explore the observability lab
05

Private Product Engineering Project

Di Torino /
NexoHab

A private operational management product combining application engineering, access control and automated regression foundations.

Private product project

How can an operational product grow with controlled access and repeatable quality checks?

The implementation

.NET 10, ASP.NET Core MVC, EF Core/SQL Server and Identity/RBAC, with operational modules and a Document Center.

Automated Regression Testing Foundation

46 tests and a GitHub Quality Gate provide a repeatable foundation for validating product changes.

Read the case study

The problem

Operational workflows need consistent authorization, document handling and confidence when the application changes.

The decision

Build on ASP.NET Core MVC with EF Core/SQL Server persistence, Identity/RBAC authorization, operational modules and a Document Center. Establish automated regression testing and a GitHub Quality Gate as part of product engineering.

The boundary

This is a private product engineering project. Source code, internal operational information and resident data are not publicly shared. The test count describes the regression foundation, not complete product coverage.

02 / Engineering principles

The result matters.
The reasoning matters too.

01

Start with risk.

Prioritize what can hurt the product or its users. Meaningful coverage comes before impressive test counts.

02

Keep the evidence.

Make failures investigable and decisions traceable. A green check alone is an incomplete explanation.

03

Keep people accountable.

Automation and AI can inform decisions. Human judgement remains responsible for accepting residual risk.

03 / About

Established practice.
Active curiosity.

I’m Javier Capa, a Senior Quality Software Engineer with 17+ years in Software Quality Assurance and Quality Engineering.

My work spans quality strategy, risk-based validation, maintainable automation, data and ETL quality, and CI quality gates. I care about making quality a shared engineering capability.

I’m also studying agentic workflows, MCP and test observability through small, executable experiments. Those projects are presented as learning labs, with findings and limitations kept visible.

Strategy & delivery
Risk-based testing · QA governance · Evidence standards
Automation & data
Playwright · TypeScript · API testing · SQL / ETL validation
Current research
OpenTelemetry · Agentic QE · MCP · Deterministic evals