Aiko Services: An open-source framework for creating awesome Machine Learning applications

In the fast moving field of Machine Learning, we are being inundated with a myriad of new frameworks and libraries, each with its own set of quirks and limitations. They might only work within a single process, or are unable to handle streaming media for multi-modal Large Language Models (LLMs), or lack support for managing multiple concurrent sessions when scaling up. This presentation introduces Aiko Services, an open-source framework designed to tackle these challenges head-on. It aims to be a flexible foundation for building scalable distributed systems. These systems will need to seamlessly integrate different technology domains that don't easily plug together ... without some serious effort ... - Diverse Machine Learning models and libraries - Streaming Media using GStreamer - Internet of Things using MQTT - Robotics using ROS2 Practical concepts will be illustrated using two small robot dogs in a live hardware demonstration. A variety of Machine Learning examples will be shown ... with their inner workings explored and explained. These examples range from object detection in video, speech translation from audio, scene description with multi-modal Large Language Models and using LLMs to mediate commands and responses between humans and the robot dogs. And time permitting, trying out some Theory of Mind experiments with these robot dogs ! It can be surprisingly easy and fun to create sophisticated ML applications ... if you just have the right software building blocks 😊