Design and implementation of outstanding enterprise software architecture which manifests services in both local and cloud computing is difficult. Historically, computer vision services ran in cloud or edge modalities but, because of huge advances in machine learning there are many use cases for a hybrid approach.
Join Tim Huckaby in elaborating his two-year journey (the successes, the failures, the lessons learned) in building an enterprise suite of software with distributed dev teams. And what the future holds for real time computer vision systems.
Typically, computer vision services that run on the edge offer incredible performance at the cost of precision, scale and flexible management. Additionally, in the current state of Machine Learning (ML) computer vision services that run on the edge can take significant expertise to build.
Computer Vision Services that run in the cloud are cost effective, easy to implement, “canned” solutions that are built by the world’s computer vision machine learning experts so all the “hard stuff” is already done for you. But, by running in a cloud modality, you sacrifice performance and require a network connection with bandwidth.
When you add the complications of providing services both in the cloud and on the edge the complexity is almost overwhelming.
Compounding the complication of this type of software architecture are containers, different types of machine learning models, different edge models, distributed computing models,
Another challenge is that managing systems like these in the cloud can get expensive quickly; especially when developers make mistakes, so a tactical plan is required to keep cloud solutions cost effective; which makes this type of cloud architecture even more challenging.
The session is designed for broad audience appeal with goals to get you smart on:
- The current state of the machine learning revolution and where it’s headed in the future
- designing, building and delivering real time machine learning based systems
- cloud / edge modalities in including the concepts of Edge, Light Edge, Heavy Edge
- software architecture for systems that use cloud and/or edge real time services.