[← back to projects]

[live]

TestFiesta

AI-powered test management platform

1 year (ongoing)

A test management tool that bends to how QA teams actually work — with an AI Copilot, versioned test history, and flat $10/user pricing.

[visit the site →]
TestFiesta screenshot 1
TestFiesta screenshot 2

The problem

QA teams were stuck managing tests in spreadsheets or paying per-feature for rigid SaaS tools that didn't integrate with their dev stack — and none of them offered intelligent assistance for actually writing test cases.

What I built

A multi-tenant SaaS platform with AI-assisted test case generation, immutable version history, bidirectional sync with GitHub, Jira, and TestRail, built-in defect tracking, and an OpenAPI-first architecture that plugs into any CI/CD pipeline — at a flat rate.

Core features

  • -AI Copilot that generates, refines, and improves test cases from plain-text requirements
  • -Versioned test cases with full history — every edit creates a new immutable version, nothing is ever overwritten
  • -Exploratory testing sessions alongside structured test runs and test plans
  • -Built-in defect tracking with bidirectional links to test executions
  • -GitHub, Jira, and TestRail sync with incremental two-way updates and tag mapping
  • -Custom fields on cases, steps, executions, and results — adapts to any team's data model
  • -Multi-tenant workspaces with role-based and fine-grained resource-level permissions

Engineering highlights

  • -Immutable versioning — each edit creates a new uid sharing a version_ref across all versions, giving teams full rollback and an audit-ready change history without extra query complexity
  • -Temporal workflow orchestration for durable multi-step operations (account provisioning, integration syncs, test run orchestration, GDPR deletion) with retry policies, tenant-scoped search attributes, and idempotency via conflict policies
  • -AWS Bedrock + LangChain with Zod schemas dynamically generated from the user's custom field config — an OutputFixingParser makes a corrective pass if structured output validation fails
  • -Relationship-based access control via OpenFGA — authorization as FGA tuples (user → role → resource), not baked into query logic, with batch authorization running as a single round-trip for list endpoints
  • -Per-tenant dedicated connection pools and schemas — a shared metadata database tracks app node and database server assignments; Temporal handles migrations and coordinated rolling deployments
  • -OpenAPI-first with TSOA — decorators generate the spec at build time; two versioned route trees (v1 legacy, trv2 current) let the API evolve without breaking existing integrations

Tech stack

Node.jsExpress.jsTypeScriptPostgreSQLKnexObjection.jsRedisAWS BedrockLangChainZodJWTPassport.jsGoogle OAuthOIDCOpenFGATemporalBullMQStripeGitHub APIJira Cloud APITestRail APIDockerKubernetesSentryVue.js

Impact

  • -Single platform replacing test case management, defect tracking, and multiple integration tools
  • -Versioned test history eliminates the "who changed this and when?" problem that plagues spreadsheet-based QA workflows
  • -AI Copilot reduces the time teams spend writing boilerplate test cases from requirements
  • -Predictable flat-rate pricing vs. per-seat-per-feature tiers from incumbents

Want something similar built?

Happy to talk it through — drop me a line.

[get in touch]