Scaleout Edge: Sovereign Edge AI orchestration and Federated Learning

In this repository, we explore model compression for transformer architectures via quantization. We specifically explore quantization aware training of the linear layers and demonstrate the performance for 8 bits, 4 bits, 2 bits and 1 bit (binary) quantization.

A camera for measuring sediment grain sizes with edge ML

Hyperdimensional computing with statistical guarantees: calibrated probabilities, conformal prediction sets, anomaly detection with a guaranteed false-positive rate

ESP32-S3 WiFi CSI Human Activity Recognition system with real-time RF sensing, ML inference, and live visualization dashboard.

An awesome list of "small but mighty" models and resources.

BEAVER automates the Edge AI lifecycle through LLM-powered orchestration. It Builds, Evolves, Analyzes, Validates, Executes, and Repairs—just like beavers in nature, but for edge devices!

Material and source code for Udacity's Introduction to TensorFlow Lite

Compress PyTorch models for edge devices — CPU-only, no GPU, no retraining. One function call.

Notes and resources from Qualcomm On-device AI course, provided by DeepLearningAI

Edge-deployable keyword spotter: INT8-quantized DS-CNN on Google Speech Commands, exported to ONNX, with fp32 vs INT8 benchmarks, a live mic demo, and a C++ inference harness.

embedded software components for event-based application development

Python ML library for person fall detection. Intended for IoT deployments with on-device inference and on-device transfer learning.

On-device TinyML on the Silicon Labs Thunderboard Sense 2 (EFR32MG12, no NPU): an 85% IMU gesture recognizer running fully on-chip at ~87.5ms, plus a deep-audited hardware reference set.

Pre-quantized models for edge inference on Cortex-M, ESP32-S3, and Raspberry Pi. Six models — keyword spotting, person detection, hand-gesture recognition, anomaly detection, voice activity detection, and binary defect classifier — each shipped as TF

Fault-tolerant Edge ML pipeline for space weather (TEC) prediction. Fuses raw ISRO telemetry with NASA APIs for sub-second, on-device inference via quantized TensorFlow Lite.

A system for monitoring statistical data distribution shifts in distributed settings

Lightweight Attention U-Net for Breast Cancer Semantic Segmentation

A curated list of machine learning models that run locally on edge devices, including phones, browsers, laptops, Raspberry Pi-class boards, Jetson, Coral, NPUs, and microcontrollers.

Edge-first XRF V2 benchmark for deployable wearable event detection (earbuds + smart-glasses), with FP/hour-calibrated metrics and reproducible run artifacts.