Speed at Scale: Optimizing The Largest CX Platform Out There
summary
Platform serves 15mi+ sessions/month and migrates a huge legacy React codebase using automated transforms, staged generations behind feature flags, and code generation to avoid a full rewrite. Migration uses JSCodeShift codemods and transformer groups; jQuery usage is detected via static analysis plus runtime proxies to report calls; polyfills load by ES module/runtime detection; modern bundle size drops 280→175 and visual metrics (Speed Index, LCP) improve by several seconds.
AI-written summary · about
chapters (10)
- 0:00Introduction and setup
- 1:48Modernizing legacy React code
- 5:21Handling multiple customer versions
- 7:30Automating code transformations
- 9:22React Suspense and side effects
- 11:09Detecting and removing jQuery usage
- 12:36Runtime detection with proxies
- 13:39Polyfilling for legacy browsers
- 15:05Performance improvements and results
- 16:12Considerations for diverse users
related talks
25mSpeed at Scale: Optimizing The Largest CX Platform Out There
13mSpeed at Scale: Optimizing The Largest CX Platform Out There
28mThe Path to Scaling a Digital Enterprise
22mPlatform-powered: Building a Frontend Platform to Scale
32mJavaScript at Scale by Stephanie Siaw (Eng) | JSConfAM19
10mHow to migrate 50,000 lines of code to React Server Components
31mSpeeding up Startup for Large React Apps
33mWeb performance in large-scale applications