App Performance Tracing

A tool for comparing per-function CPU time between release versions of an mobile application

Product: Mobile App Performance Tracing

The Problem: Meta’s iOS and Android apps (the blue app, Messenger, Instagram, Whatsapp, etc.) are updated constantly with new code from 10,000+ product engineers. New code often brings performance degradations via app size bloat and code errors. The company needed a method to compute and diagnose code-induced performance problems.

The Solution: We created a platform called Comparison View that takes function callstack traces measuring CPU time from thousands of devices, normalizes the data, and computes the performance (speed) gains and losses between versions of a given application. The tool allows engineers to perform root-cause analysis on app slowdowns, by pinpointing function-level code changes downt to the contributing commits.

(Watch this quick video walkthrough of the product)

My Role at Meta

Full time employee

Lead user research, UX design, and interaction for multiple products, including:

  • Application performance stack tracing and visualization
  • Experimentation management
  • Natural language processing
  • Engineering release management tools
  • Productivity tools (task management, code review)
App Performance Tracing | Paul Van Slembrouck

App Performance Tracing

A tool for comparing per-function CPU time between release versions of an mobile application

Product: Mobile App Performance Tracing

The Problem: Meta’s iOS and Android apps (the blue app, Messenger, Instagram, Whatsapp, etc.) are updated constantly with new code from 10,000+ product engineers. New code often brings performance degradations via app size bloat and code errors. The company needed a method to compute and diagnose code-induced performance problems.

The Solution: We created a platform called Comparison View that takes function callstack traces measuring CPU time from thousands of devices, normalizes the data, and computes the performance (speed) gains and losses between versions of a given application. The tool allows engineers to perform root-cause analysis on app slowdowns, by pinpointing function-level code changes downt to the contributing commits.

(Watch this quick video walkthrough of the product)

My Role at Meta

Full time employee

Lead user research, UX design, and interaction for multiple products, including:

  • Application performance stack tracing and visualization
  • Experimentation management
  • Natural language processing
  • Engineering release management tools
  • Productivity tools (task management, code review)