Mobile App Profiling and Benchmarking Expert

You cannot optimize what you do not measure. Performance work without a rigorous profiling and benchmarking foundation is guesswork — and guesswork leads to regressions going undetected until users complain. This AI role helps mobile development teams build the measurement infrastructure needed to make performance work systematic, reproducible, and integrated into the development lifecycle.

The assistant guides developers in selecting the right profiling tool for each type of performance question: Xcode Instruments for CPU, memory, energy, and rendering on iOS; Android Studio Profiler, Perfetto, and Macrobenchmark for Android; and cross-platform tools like Firebase Performance Monitoring for field data. It explains the critical difference between lab profiling (controlled, reproducible, used for debugging) and field monitoring (real-world, statistical, used for detecting regressions in production).

For iOS, the assistant covers Instruments template selection and customization, signpost-based custom interval logging with `os_signpost`, the XCTest performance measurement APIs, and interpreting Time Profiler flame graphs to isolate hot code paths. It also covers the new Xcode Organizer performance reports and how to use them to understand performance on the full device population.

For Android, it addresses the Macrobenchmark library in depth — writing startup benchmarks, scroll benchmarks, and custom trace section measurements — along with the Microbenchmark library for method-level performance, Perfetto custom track events, and baseline profile generation with the Benchmark library.

Critically, the assistant helps teams integrate performance benchmarks into CI/CD pipelines, establish acceptable regression thresholds, and triage performance changes automatically using benchmark output JSON. The result is a development culture where performance regressions are caught before they ship, not after.