Hands-on Train and Deploy ML
Build a Serverless API to predict crypto prices in 10 steps
In this tutorial you won't build an ML system that will make you rich. But you will master the MLOps frameworks and tools you need to build ML systems that, together with tons of experimentation, can take you there.
With this hands-on tutorial, I want to help you grow as an ML engineer and go beyond notebooks.
1. Create the virtual environment with Python Poetry
👨🏽💻Source code | 🎬 Video lecture
Let's set how easy it is to develop and package your Python code using Poetry.
2. Generate training data
👨🏽💻Source code | 🎬 Video lecture
Without training data there is no ML.
In this lecture you will learn how to generate a basic training dataset from the Coinbase Historical data API.
3. Build a baseline model
👨🏽💻Source code | 🎬 Video lecture
Before using any ML algorithms you need to establish a baseline performance on the test set. A rule-based model, for example a moving-average, can give you a solid baseline for financial prediction problems like this.
4. Build ML Models
👨🏽💻Source code | 🎬 Video lecture
In this lecture you will how to track experiments and speed up your development cycle using CometML.
5. Deploy the model as a Serverless REST API
👨🏽💻Source code | 🎬 Video lecture
Let's deploy our ML model as a REST API using the Serverless platform Cerebrium.
Forget about Docker, IAM roles, and EC2 instances. Serverless ML is about focusing on what differentiates your ML product, not setting up and mantaining infrastructure.
