bonolive ~/portfolio
PT
Back to all projects

AI-Driven Test Automation Framework

A framework that generates and runs UI test cases automatically, using computer vision to drive real Android devices.

Problem
Manual regression testing was slow, repetitive and the bottleneck before every release.
Solution
A Python framework that auto-generates test cases and validates UI with computer vision + UIAutomator, orchestrated with Docker and Jenkins.
Impact
Automated 250+ test cases, saved 100+ developer hours and cut manual testing time by 40%.
Stack
  • Python
  • Computer Vision
  • Docker
  • Jenkins
  • RabbitMQ
  • UIAutomator

Context

Regression testing was done by hand before each release — slow, error-prone and the single biggest bottleneck in the delivery pipeline.

Architecture

A Python framework generates test cases and validates screens using computer vision combined with UIAutomator to drive real Android devices. Jobs are distributed through RabbitMQ queues and run in Docker containers, wired into Jenkins so the suite fires on every build.

Details

  • Computer vision asserts on what the user actually sees, catching visual regressions text assertions miss.
  • The queue-based design scales horizontally across multiple devices.
  • Results feed back into CI, blocking merges on real failures.

Outcome

The suite grew past 250 automated cases, saved 100+ developer hours, and reduced manual testing time by 40% — turning releases from a careful ritual into a routine.