Explore production applications, edge AI systems, and academic research battle-hardened against resource exhaustion, thermal throttles, and quantization drift.
The official cross-platform reference demo application running on Android, iOS, and Web. Demonstrates live fault injection toggles, step-by-step interactive onboarding tutorial, real-time latency graphs, and instant SQLite report exports.
A high-throughput runtime observability suite for quantized 4-bit and 8-bit Large Language Models running on consumer edge devices. Integrates SATE AI adaptive chaos schedules to profile token generation degradation under aggressive thermal states.
Autonomous optical inspection system deployed on edge manufacturing tablets. Leverages SATE AI Malformed Input and Quantization Drift injectors to test sensor glitch resilience and camera pipeline corruptions.
An offline-first clinical assistant running skin lesion classification and respiratory sound analysis on low-power devices in remote regions. Tested using SATE AI Confidence Threshold and Memory Pressure injectors.
In-browser chaos harness for WebAssembly and WebGPU neural runtimes. Injects synthetic WebGL buffer stalls, context losses, and network latency drops directly into browser inference pipelines.
Are you using SATE AI to test and harden your on-device models? Showcase your work to thousands of mobile and edge AI engineers worldwide.
Execute SATE AI against your on-device model and generate an HTML or JSON report via the CLI or Flutter SDK.
Prepare your project name, 2-3 sentence description, links (GitHub/website/store), a screenshot or GIF, and a brief testimonial.
Open a PR adding your card to docs/showcase.html or file an issue using the Showcase Submission template.
Add the verified SATE AI Tested badge to your repository README to show users that your AI is rigorously validated.
Copy and paste this markdown into your project README: