Scale QA Automation for ChatGPT

Project Background

OpenAI's QA Engineering team needs to expand automated test coverage for a new ChatGPT web release that adds conversation export, improved file upload flows, and updated model switching behavior. The release is scheduled in 8 weeks and will be exposed to millions of users, so leadership wants stronger regression protection without slowing the launch.

You are the QA Engineer responsible for driving the execution plan for test automation across the ChatGPT web app and key API-backed workflows. The working team includes 4 frontend engineers, 3 backend engineers, 2 QA engineers, 1 product manager, and 1 design lead. The project is urgent because the previous release caused 3 production regressions that escaped manual testing.

Key Stakeholders

Engineering wants fast CI feedback and minimal maintenance overhead. Product wants all launch-critical user journeys covered before code freeze. The release manager wants a clear go/no-go process, while the infrastructure team is concerned about longer test runtimes in CI. These priorities conflict: broader coverage increases confidence, but can delay builds and launch readiness.

Constraints

• Timeline: 8 weeks until launch, with code freeze at the end of Week 6
• Budget: $45,000 for tooling, test environment improvements, and contractor support
• Team capacity: 2 QA engineers, with one shared 30% on another release
• CI budget: test suite cannot add more than 12 minutes to the main pipeline
• Dependencies: stable staging environment by Week 2, telemetry hooks by Week 3, and final product requirements by Week 2

Complications

1. The staging environment has intermittent auth failures that make end-to-end tests flaky.
2. Product is still debating whether file upload limits will change, which affects expected test coverage.
3. A senior frontend engineer is pushing for mostly unit tests, while the release manager wants end-to-end launch gates.

Your Task

1. Define a practical automation strategy and roadmap for the 8-week release.
2. Prioritize which ChatGPT user flows must be automated before launch versus deferred.
3. Propose how to manage flaky tests, CI runtime limits, and changing requirements.
4. Specify launch readiness criteria, rollback triggers, and post-launch monitoring.
5. Identify the top execution risks and how you would mitigate them.