Building AI Solutions on the Edge Training in Thimphu, Bhutan

This training introduces participants to the concepts, tools, and best practices for designing and deploying AI solutions on edge devices. It focuses on practical skills for creating intelligent systems that run AI models locally on devices such as Raspberry Pi, NVIDIA Jetson, and smartphones—enabling real-time, secure, and low-latency AI applications.

Fundamentals and Setup of Edge AI Solutions

Introduction to Edge AI Solution Design

• What is Edge AI?
• Why build AI at the edge (speed, privacy, offline, cost)
• Overview of common use cases: smart cameras, sensors, mobile apps

Edge AI Architecture and Components

• Hardware overview: Raspberry Pi, Jetson Nano, Coral, mobile phones
• AI development tools and model formats: TFLite, ONNX, OpenVINO
• Deployment pipeline: training → optimization → inference

Model Optimization Techniques

• Reducing model size with quantization and pruning
• Converting models to run on edge devices

Building Real-World Edge AI Applications

Building AI Applications with Edge Inputs

• Integrating camera, microphone, or sensor input
• Example pipelines: real-time object detection, voice command recognition
• Handling real-time constraints: latency, buffering, output display

Use Case Development and Edge Deployment

• Choose a real-world scenario: security camera, smart assistant, environmental monitoring
• Design full solution: device + AI model + input/output
• Tools: Edge Impulse (optional), NVIDIA Jetson SDK, OpenCV

Final Review and Next Steps

• Common challenges and how to solve them (power, updates, scalability)
• Resources for advancing in edge AI
• Wrap-up Q&A and feedback