Leveraging Elixir for Scalable Embedded AI Point of Sale Systems

Alfonso Gonzalez shares his team's journey in developing a real-time, AI-based, unattended point of sales system using the Elixir ecosystem. The core components utilized included LiveView for real-time dashboards, Phoenix for extending the application's core with websockets, and Nerves for embedded firmware development. The initial approach faced hardware limitations on performance and upgradeability, which led them to adapt their system with Elixir's concurrency capabilities and the use of Nerves to improve reliability and maintainability. Integration with Amazon Web Services (AWS) IoT and Greengrass provided secure and efficient remote updates, authentication, and device management. Gonzalez details the process of building custom Linux systems with Buildroot to improve media handling capabilities essential for their AI models. Towards the end, he touches on using private 5G for network management to support mass service scalability. This talk showcases the versatility and robustness of the Elixir ecosystem when creating complex embedded systems for commercial retail applications.