On this weblog sequence, I’ll introduce you to MLOps, combining each theoretical insights and sensible steering. The purpose will not be solely to discover the important thing ideas and applied sciences within the area but additionally that will help you apply them successfully in your personal tasks and workflows.
The primary sources behind this sequence are:
- Sensible MLOps: Operationalizing Machine Studying Fashions (1st Version) by Noah Reward and Alfredo Deza Printed By Oreilly
— A extremely advisable learn for a deeper understanding;
— I used some(not all) of the construction of my subjects and titles just like the ebook due to coherency - My self-study and coursework in DevOps, together with hands-on expertise from private and tutorial tasks.
- Printed articles and sources on machine studying and cloud computing and college programs.
Be aware that I write these blogs for many who are not less than a newbie to DevOps which means you could have somewhat understanding of it’s subjects and wish to be taught the ML aspect of issues and finally MLOps, so I’ll not totally clarify foundational subjects like Bash scripting or CI/CD pipelines however they are going to be used closely within the examples.
I additionally encourage you to ask chatGPT(or every other desired AI) or lookup any subject that you could be want extra explaining for.
Let’s start the primary a part of the sequence:
This primary half makes use of most of it’s sources and content material from the talked about ebook for the reason that first chapter is offered at no cost and it does a great introduction to the subject.
Cloud Native ML PLatforms
- AWS SageMaker
- Azure ML Studio
- GCP AI Platform
Containerized Workflows
- Docker format containers
- Kubernetes
- Personal and public container registries
Serverless Applied sciences
- AWS Lambda
- AWS Athena
- Google Cloud Features
- Azure Features
Specialised {Hardware} For Machine Studying
- GPUs
- Google TPU (TensorFlow Processing Unit)
- Apple A14
- AWS Inferentia Elastic Inference
Huge Information Platforms and Instruments
- Databricks
- Hadoop/Spark
- SnowFlake
- Amazon EMR (Elastic Map Cut back)
- Google Huge Question
Usually one clear sample about machine studying is how deeply it’s tied to cloud computing. It’s because the uncooked substances of machine studying occur to require large compute, intensive information, and specialised {hardware}.
