AI/ML at the Extreme Edge with WebAssembly: A Path Forward
summary
Claim: WebAssembly plus CRDTs make AI/ML practical on constrained, disconnected devices by running models and sync logic locally. Demonstrated by compiling TensorFlow compiled to WebAssembly and running an on‑machine anomaly detector for factories where outages cost ~$100k/hour. Uses Wasmtime/WASI runtimes, vector clocks for merges, and an open‑source WASM‑based CRDT engine that ensures CRDTs guarantee conflict-free convergence. Result: inference and local learning run at near-native (≈90%) performance and avoid fragile cloud dependencies.
AI-written summary · about
chapters (5)
- 0:00Start
- 9:43The End of Cloud Computing
- 19:15Key Properties of Web Assembly
- 21:33Machine Anomaly Detection
- 26:15Conflict-Free Replicated Data Types
related talks
33mWebAssembly based AI as a Service with Kubernetes
27mAI inference on the Edge cloud using WebAssembly
41mWebAssembly Based AI as a Service
01h 04mDevelop, Build & Deploy Lightweight AI Services with WebAssembly
37mGetting Started with AI and WebAssembly
39mWebAssembly-Powered Machine Learning
40mWebAssembly for AI Infra : A Lightweight, Fast, and Secure Alternative Approach
32mEmpower WebAssembly ML Over Diverse AI Hardware Accelerators Outside the Browser