Android to Polyglot Backend Dev
summary
Moves from Kotlin UI work to spinning up TypeScript and Python backends using Cursor, Claude and ChatGPT, then rewrites TS features into Python after an acquisition within months. Uses agents.mmd file to encode repo build/test/commit conventions, types ultrathink prompts for deeper code generation, and trains agents to run CI checks, rebase branches, fix merge conflicts and open PRs. Builds a working Python server for the build-upload endpoint; uses AI to scaffold tests, write load-test scripts that submit many APKs, and debug CI failures.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 2:10Career shift Android to backend
- 6:28Getting started with AI agents
- 8:56Double-checking AI code and tests
- 10:42Implementing features with AI
- 13:13Writing and running tests
- 15:46Automating git workflows
- 18:13Diagnosing CI failures
- 20:19MCP for docs and tools
- 21:44Planning issues with AI
- 23:29Debugging with Sentry
- 24:58Managing AI agent context
- 26:21Running multiple agents
- 28:07Summary
related talks
41mHow AI Will Change Mobile Development
38mOne Android Dev, Ten AI Teammates: Scaling Delivery With Claude Code
14mA Song of Types and Agents
34mMulti-coding agent workflows
50mSpec-driven Development using Coding Agents
37mDevelop faster with Android Studio, AI and skills
01h 00mBuilding a Polyglot Multi-Agent AI Architecture
20mAI in Android Development at Duolingo