Speaker Details

Robbe Sneyders

Robbe is the Head of Delivery at ML6, a machine learning services company founded in Belgium with international offices throughout Europe. He leads the technical team and is responsible for the successful and high quality delivery of their projects. Robbe considers himself a machine learning generalist, but has especially deep knowledge about machine learning systems design with a focus on representation learning based systems and the MLOps methodology.

Robbe is a big fan of open source. He's a maintainer and leading developer open source Connexion API framework and contributes to the wider Python API ecosystem. He made some major contributions to Apache Beam, on which he delivered a keynote at the European Beam summit. He's also been featured as a speaker on Tensorflow meetups, Google Cloud meetups, and ML conferences across Europe.

Connexion: API-first for all

Connexion is a Python API-first framework that automagically handles HTTP requests based on an OpenAPI specification (~Swagger spec) of your API. Connexion starts from the API contract, maps the endpoints to your Python functions, and provides features such as security, serialization & validation, all automatically configured based on the contract. This makes it unique, as many tools generate the specification based on your code instead. You can describe your REST API in as much detail as you want and Connexion guarantees that it will work as you specified. Connexion was originally developed by Zalando and recently released to the community. Previously, Connexion was built on top of Flask, but as of version 3.0, it works with almost any Python API framework by leveraging the standard ASGI interface.

In this talk we'll discuss:

  • API-first as a development practice
  • The Connexion framework and features
  • ASGI middleware based architecture

Apache Beam is an open-source, unified programming model for batch and streaming data processing pipelines that provides a variety of language SDKs such as Python, Java, Go, and more. This enables the creation of multi-language pipelines where each step can be implemented in the most suitable language. This enables collaboration between teams with different language preferences, and across domains with different default languages, such as Java for data processing and Python for ML modeling.

In this talk we'll look at the Apache Beam framework and how to use it to orchestrate batch and streaming ML workflows with Python and Java components.