PyData Montreal Meetup #23 (in-person | en personne)

IMPORTANT: To access the meetup:

• The entrance is located at 1100 Avenue des Canadiens AND NOT on Rue Saint-Antoine (as some GPS may indicate). Once you enter the building, enter the elevator in front of you, and we will be waiting for you on the 4th floor.
• To access the room, there is some security measure (give some identifications), so you can fill out this form to enter your information early to facilitate the access to the room

Welcome to the first meetup of 2023! We're thrilled to host this event at the Coveo office in Montreal. This meetup will focus on two of the most exciting and rapidly-evolving fields: deep learning and data.

Deep learning has revolutionized how we approach problems, allowing us to build intelligent systems that can learn from large datasets and make predictions with incredible accuracy. At the same time, the amount of data generated by businesses and individuals has grown exponentially, creating new challenges and opportunities for researchers, engineers, and analysts alike.

At this meetup, you will have the opportunity to learn from local professionals in both fields. So join us for an evening of networking, learning, and fun.

More details on the event are below 👇👇👇

Agenda
5:30 pm - Doors open
6:00 pm - Introduction
6:10 pm - “Deep learning in production at Coveo” by Francis Piché & Alexander Bustamante Cubas, Coveo
6:50 pm - Break and networking (drinks and pizzas will be provided)
7:10 pm - "How to scale user requirements in your data platform?" by Matthieu Delacour, Coinlist
7:50 pm - Final notes, break, and networking

Talks
Deep learning in production at Coveo
Abstract: In this talk, we present DELI (DEep Learning Infrastructure), the system put in place by the ML Platform team at Coveo to enable ML teams to efficiently take deep learning algorithms from inception to production on a variety of use cases. The general design and challenges are expected to be discussed and how the applied scientists leverage DELI to implement data preparation, training, and serving by using tools such as Pytorch, Raytune, and Pytorch Lightning.
About the speakers
Francis is a Senior Software Developer at Coveo; he started his career working in developer tooling and quickly switched to building machine learning platforms. The love for CI/CD systems and automation has stayed with him as he transitioned towards a more formal MLOps role, where he now focuses on applying CI/CD practices to an ML context at scale. Francis is particularly interested in the human aspect of software development and how the tools we build and choose affect the developer experience.
Alexander is an applied scientist focusing on Natural Language Processing (NLP) applications. With more than 14 years of experience in software development, he refocused his career on Machine Learning/Deep Learning a few years ago by completing a Master's program at the University of Montreal (UdeM) supervised by Mila (Quebec Institute of Artificial Intelligence). He quickly developed a strong interest in leveraging Data Analysis and NLP to build robust applications serving thousands of users in production at scale. He's passionate about simple solutions to complex problems, a principle he applies to analytical and technical tasks.

Codebase organization: How to be prepared for the unknown?
Abstract: Lots of us start with a mono repo that becomes unmanageable. Over time your users have different requirements (DBT, ML, heavy batch pipeline, web scraping, etc.), and the maintenance of that mono repo becomes horrible. Based on my experience at Shopify and CoinList, I will share how I think about that and how you can ensure you won't end up in this situation from day 1.
About the speaker
Matthieu is an Engineering autodidact; he started his career in Data Science. To him, the best part was to build platforms and services at scale. With his colleagues, he built the new ML platform of Shopify (Merlin), leading him to become a code maintainer of Feast (ML feature store). A year ago, He joined CoinList to build their data team and their data platform from scratch.