Flama (https://flama.dev/) is a Python library that provides a standard framework for developing and deploying APIs with a special focus on machine learning (ML). With Flama, developers can build highly performant GraphQL, REST, and ML APIs with a ridiculously simple deployment process that can be as simple as a single line of code. Flama provides an easy-to-learn philosophy that simplifies the building of asynchronous and production-ready services. Flama is an ideal solution for developers who want to automate ML deployments.

Workshop Overview

This workshop will provide an introduction to Flama, showing attendees how to set up a standard development configuration, and to design a basic ML pipeline to mimic a prototypical ML project. Participants will learn how to pack the ML project with Flama, and serve the model with a single command line. The workshop will highlight the advantages of using Flama to build ML APIs, and show attendees how to serve multiple models just using configuration files. Finally, participants will learn how to serve customized ML APIs.

Workshop Outline

The workshop will be divided into six parts:

Part 1: Introduction to Flama

In this section, we'll provide an overview of Flama, explaining its main features and how it works. Participants will learn how to set up their development environment, and we'll provide examples of how Flama can simplify the development of ML applications.

Part 2: ML Pipeline Design

In this section, we'll teach attendees how to design a basic ML pipeline. We'll cover topics such as data pre-processing, model training, and model evaluation, and provide examples of how to implement each step in Python.

Part 3: Packing the ML Model Pipeline with Flama

In this section, we'll show attendees how to pack their ML pipeline with Flama, making it easy to manage and version their models. We'll explain how Flama uses OpenAPI to define the API specification, and how to use Flama's built-in methods to safeguard rich ML artifacts.

Part 4: Serving the ML Model with Flama CLI

In this section, we'll show attendees how to serve their ML model with Flama CLI using a single command line. We'll explain how to use the Flama CLI to start the API server and how to test the API endpoints.

Part 5: Serving Multiple Models with Flama CLI

In this section, we'll show attendees how to serve multiple models in the same ML API with Flama CLI. We'll explain how to use configuration files to specify the models to be served, and how to interact with all of them.

Part 6: Customized ML APIs with Flama

In this section, we'll show attendees how to serve customized ML APIs with Flama, adding custom methods and endpoints. We'll explain how to use Flama's component system to inject dependencies and how to create custom endpoints.

By the end of this workshop, attendees will have a good understanding of Flama and how it can be used to simplify the development and deployment of ML applications.