Beyond Notebooks - Serverless Machine Learning
Overview
The "Beyond Notebooks - Serverless Machine Learning" course is designed to simplify the process of building intelligent systems using machine learning (ML) models. Traditionally, developing such systems required expertise in cloud computing or managing infrastructure, which could be a daunting task. This course introduces a serverless approach, where participants only need to know how to write Python programs. These programs can be scheduled to run as pipelines, and the produced features and models are managed through a serverless feature store or model registry. This course is perfect for those who want to incorporate ML into practical applications without the overhead of managing underlying systems.
Prerequisites
To embark on this learning journey, individuals need a basic understanding of Python, familiarity with Pandas (a data manipulation library), and an active GitHub account.
Modules
The course is divided into comprehensive modules, each focusing on different aspects of serverless ML:
-
Module 00 - Introduction: Offers introductory content, including the importance of serverless ML and setting up a development environment.
-
Module 01 - Pandas and ML Pipelines in Python: Guides learners in writing their first serverless app using Pandas and ML pipelines.
-
Module 02 - Data Modeling and the Feature Store: Explores data modeling with feature stores, specifically through a credit-card fraud prediction service.
-
Module 03 - Training and Inference Pipelines: Covers the implementation of training and inference pipelines along with the model registry.
-
Module 04 - Serverless User Interfaces: Demonstrates how to create user interfaces for ML systems using simple Python and HTML.
-
Module 05 - MLOps Principles and Practices: Introduces the fundamentals of MLOps, focusing on best practices like versioning and data validation.
-
Module 06 - Operational Machine Learning Systems: Concentrates on real-time ML and operational practices in serverless environments.
Learning Outcomes
By completing this course, participants will learn to:
- Develop and operate AI-enabled prediction services on serverless infrastructure.
- Deploy and manage features and models on serverless setups.
- Utilize training and inference pipelines for model training.
- Design serverless user interfaces for ML services.
- Understand MLOps essentials, including versioning, testing, and monitoring.
- Implement real-time serverless machine learning systems.
Course Content
The course encompasses a wide range of topics, such as creating serverless apps using Pandas, feature engineering, running batch and inference pipelines, developing user interfaces using tools like Gradio, Streamlit, and automated testing and versioning practices.
Target Audience
This course is ideal for individuals with a basic understanding of ML and Python, looking to move beyond static datasets and deploy predictive services efficiently. It caters to professionals seeking to demonstrate ML models' value in practical settings and those wishing to integrate ML capabilities into existing applications.
Why Choose This Course?
This course stands out because it requires no advanced operations experience. Participants will learn the essentials of MLOps and work with both raw and live data, with no additional costs involved, thanks to services like GitHub Actions and Hopsworks.
Requirements
Participants need:
- Python environment with a notebook setup (Jupyter or Colaboratory).
- Active GitHub and Hopsworks accounts.
Key Technologies
- Development Environment: Focuses on Python programming within notebooks.
- GitHub: Used for code management and running workflows.
- Hopsworks: Provides a platform for feature management with a free storage tier.
Register now for the Serverless ML Course and embark on a journey to become a serverless machine learning engineer without incurring costs on running pipelines or managing features/models/user interfaces. The course is self-paced, allowing flexibility to learn and implement serverless machine learning systems.