Engineering Culture by InfoQ

Using AI Code Generation to Migrate 20000 Tests

25 snips
Jun 6, 2025
Guest
Sergii Gorbachov
Sergii Gorbachov, a Staff Engineer at Slack, shares insights on a groundbreaking project that migrated 20,000 tests in just 10 months. He discusses the blend of AI and traditional coding techniques, emphasizing that AI alone wasn't sufficient without human oversight. The conversation highlights the evolving role of developers as they collaborate with AI to tackle complex tasks and ensure code quality. Sergii also touches on the challenges of pull request dynamics, the impact of testing metrics, and how generative AI can be harnessed for specific engineering problems.
Ask episode
AI Snips
Chapters
Transcript
Episode notes
ANECDOTE

Slack's 20,000 Test Migration

  • Sergei shared how Slack migrated 20,000 tests in 10 months using AI and traditional tech combined.
  • The project sped up significantly, avoiding what would've been 10-15,000 engineering hours manually.
INSIGHT

AI Complements, Not Replaces

  • AI alone was overhyped and insufficient for code conversion tasks.
  • Traditional coding tools remain critical; AI only complements but cannot replace them fully.
INSIGHT

Developers as AI Co-authors

  • Developer roles shift from authoring code to co-authoring with AI and primarily reviewing and validating AI-generated code.
  • AI handles straightforward coding, freeing developers for complex problem-solving and control.
Get the Snipd Podcast app to discover more snips from this episode
Get the app