test.describe('Bruno Bacelar', () =>
AI Solutions Engineer. Twenty years in software, the last eight building the test frameworks that catch what breaks before your users do. Now I build the AI agents, automations, and production systems that replace manual work entirely.
Running 7 tests using 1 worker
● renders hero without layout shift
test.describe('selected work')
Five engagements. Hover a card to see its trace: how each project moved from audit to green.
MCAP · Framework rearchitecture
passedCanada's largest independent mortgage finance company had a test suite splitting at the seams across lender brands. I rearchitected it into a single multi-lender Cypress framework: shared specs, per-lender configuration, hundreds of tests stabilized and running in Azure DevOps on every merge.
GLE · Built from zero
passedGreenfield API automation in .NET 8 with NUnit and xUnit, plus custom Roslyn analyzers that catch violations of the team's own conventions at compile time. The framework reviews your code before your reviewer does.
QA Pilot · AI agent, self-directed
runningA test case generator that reads your Jira tickets, Azure DevOps work items and GitHub pull requests, then drafts structured test cases on the Claude API. Built because every team I have joined had the same backlog of untested acceptance criteria.
Diário de Bordo · Independent project
passedA live app that helps families plan a move to Canada. The financial estimate feature blends AI-generated market pricing with deterministic, hand-coded government fee logic, so the numbers that matter most are never left to a model's guess. Real users, live today.
Olea Connects · Independent project
runningFull technical build for a nonprofit governance platform: Next.js, Supabase, auth, payments, and member management. Integrations to Attio and QuickBooks run on versioned outbox workers, so a failed sync retries safely instead of quietly losing data. Launching mid-August.
expect(stack).toBeBattleTested()
No percentage bars. A capability either survives production or it does not.
E2E automation
Playwright · Cypress
8 years, from first spec to framework governance
✓ pass
Languages
TypeScript · C# · SQL
Typed everywhere. Untyped test code is a bug factory
✓ pass
API testing
REST · NUnit · xUnit · Postman
Contract-first, environment-agnostic
✓ pass
CI/CD
Azure DevOps · GitHub Actions
Pipelines with quality gates, not decoration
✓ pass
AI agents & automation
Claude API · OpenRouter · n8n · MCP-style connectors
Agents that read, decide, and act, with a human checking the risky parts
✓ pass
Mobile
Appium · BDD / Gherkin
Earlier chapter, still in the toolbox
✓ pass
beforeAll('the person')
I wrote my first line of production code in Brazil more than two decades ago. Since then I have been a developer, a consultant, and for the last eight years the person teams call when their test suite is flaky, slow, or missing entirely. These days I spend most of that instinct on AI agents and automation instead, catching what an agent gets wrong before it ships, not just what a human gets wrong.
I am based in Canada, work in English and Portuguese, and I am at my best in the gap between building fast and building something that still works in a year: real testing, real reliability, not a demo that happens to work once.
Outside of work I write psychological thrillers under a pen name. It turns out that hunting plot holes and hunting bugs are the same skill, and I cannot switch either one off.
test('hire bruno', async () => ...)
I take on AI agent and automation engineering, integration work, and advisory on bringing AI into real production workflows. English or Portuguese, remote-first.