MLOps at Edge Analytics | Introduction
At Edge Analytics, we strive to develop machine learning applications that are transparent and reproducible. Machine learning projects are often composed of many parts, and the rapidly available software solutions for managing these parts are collectively called Machine Learning Operations (MLOps).
We’ve found investing in MLOps infrastructure accelerates our ability to extract meaningful insights for the problems we solve. A robust MLOps infrastructure optimizes our iteration time and increases our trust in the models we develop for clients. In this blog series, we’ll discuss an example MLOps pipeline using some of our favorite tools.
How we use MLOps at Edge Analytics
Our team develops and deploys ML models to both edge devices and the cloud. Tracking model development and deployment is critical to ensure algorithms behave as expected in the wild. We’ll demonstrate one example of a full pipeline using tools that have worked well for our team. The example is a modular architecture with functionality broken into five major steps. These steps include:
The specific structures of MLOps pipelines will vary from project to project. At Edge Analytics, we use the best tools for the job at hand; however, we consistently use the above five points as guiding pillars for MLOps pipeline development.
Considerations for pipeline development
Building an MLOps pipeline comes with countless trade-offs for balancing structure and flexibility. In selecting the best tools for our clients, we are guided by two major principles:
- Code abstractions for third-party tools should be simple, consistent, and well documented.
- No single platform has the best of all solutions, and new features are frequently available. We should maintain a flexible pipeline capable of interchanging third-party platforms.
There are several services that provide end-to-end MLOps support, such as AWS SageMaker and Ray. However, we aim to avoid tight coupling with any given service by enabling integration with a variety of Python packages and services.
Lastly, it should be acknowledged that there is no one way to build an MLOps pipeline. There are many incredible tools that can assist in developing and tracking ML models. We hope this example pipeline gives you a good place to start and encourages you to seek out the methods that best serve your project!
A blog series to showcase our process
Over this series of blog posts, we will examine each of the five central MLOps pipeline pillars more closely in the context of an example project classifying white blood cell images by cell type. And with that, let’s take a look at data storage…
Machine learning at Edge Analytics
Edge Analytics helps companies build MLOps solutions for their specific use cases. More broadly, we specialize in data science, machine learning, and algorithm development both on the edge and in the cloud. We provide end-to-end support throughout a product’s lifecycle, from quick exploratory prototypes to production-level AI/ML algorithms. We partner with our clients, who range from Fortune 500 companies to innovative startups, to turn their ideas into reality. Have a hard problem in mind? Get in touch at info@edgeanalytics.io.
