Rust At The Edge
summary
Moves content recommendation into the edge using Rust + WebAssembly and a serverless compute platform, preprocessing the Met Museum CSV into vectors with a Rust tool, packing those into 500 cluster files in a KV store, computing a median vector for viewed IDs, and searching the relevant cluster to return recommendations in about 34 ms median recommendation; cold-start responses show ~11 ms, every request runs in an isolated WASM instance, and the demo deploys and scales without bespoke infra.
AI-written summary · about
chapters (10)
- 0:00Introduction to edge computing
- 1:30Comparing deployment models
- 5:13Why WebAssembly at the edge
- 6:53Defining the edge
- 10:42Deploying Rust to the edge
- 14:10Edge computing vision
- 16:42Machine learning recommendation engine
- 24:26Implementing the recommender system
- 31:27Edge gaming demonstration
- 34:04Q&A and concluding thoughts
related talks
25mWTF, Rust for Web Development??
30mImproving web app with Rust and WASM
41mSpeed, Safety, Sustainability. Pick 3 with Rust & Serverless
23mRust in Edge Computing: Enhance Performance & Security
42mRust, WebAssembly, and the future of Serverless
38mRust on the frontend
26mJavaScript Won the Web, Rust Is Taking the Critical Path
47mRewrite the web with Rust and Web Assembly