eZansiEdgeAI/.github

Small-footprint AI solutions for edge devices — with African flavour # eZansiEdgeAI 🌍⚡ **Small-footprint AI solutions for edge devices — with African flavour.** --- ## 🚀 Vision eZansiEdgeAI is an initiative to design and deploy **lightweight, containerised AI stacks** that run efficiently on edge devices such as the Raspberry Pi. Our goal is to make AI accessible, portable, and adaptable — without sacrificing performance. We believe in: - **Ncane (small)**: Compact models and stacks that fit resource-constrained devices. - **Shesha (fast)**: Optimised inference for real-time edge applications. - **Khanya (light)**: Illuminating insights with minimal overhead. - **Umngcele (edge)**: Bringing intelligence to the edge of networks and communities. --- ## 🧩 Approach We are building **modular AI stacks** packaged in Linux containers. This allows developers to: - Develop and train on powerful machines (e.g., Mac mini, Linux workstations). - Package AI stacks into Docker/Podman containers. - Deploy seamlessly onto Raspberry Pi or similar edge devices. - Swap or update stacks easily for different scenarios (vision, speech, anomaly detection, etc.). --- ## 🛠️ Technology Choices - **Containerisation**: Docker and Podman (rootless Podman preferred for Pi). - **Lightweight OS**: DietPi and Armbian are under evaluation for best performance and stability. - **Hardware Targets**: Raspberry Pi, ARM-based edge devices, Mac mini for development. - **Scalability**: Multiple curated stacks for different AI concepts, easily pulled and deployed. --- ## 📂 Organisation Structure This GitHub organisation will host multiple repositories, each representing a curated AI stack: - `vision-stack` 👁️ — Computer vision models for edge devices. - `speech-stack` 🎙️ — Lightweight speech recognition and synthesis. - `anomaly-stack` 📊 — Edge-based anomaly detection for IoT. - `utils` 🛠️ — Shared utilities, scripts, and deployment helpers. Future stacks will be added as the community grows. --- ## 🤝 Collaboration This project was born from conversations betwe …