MLOps.community Berlin Meetup 03

Hello Community!

On February 2nd we will have the third MLOps.community meetup in Berlin \0/. It will consist of 2 main talks, a couple of lighting talks and time to socialize (food and drinks included) :). The event will take place in Google's office in Tucholskystraße 2 10117 Berlin - where we have a capacity limit of 150 people so please confirm your seat in advance.

This time we are hosting 2 main talks.

It's a pleasure to receive Steffen Klempau - he will be talking about a hackathon at scale he organized. Steffen is a MLE at Capgemini in Berlin. He has worked in different roles at various clients for Capgemini and, prior to that, as a data engineer in a mid-sized company. He was the lead architect for the GDSC and was the one constantly telling everyone they won’t get a bigger GPU.

About Steffen's talk: The Global Data Science Challenge (GDSC) is an annual event hosted by Capgemini. The goal of the 2022 challenge was helping to fight river blindness which is affecting millions of people. To combat the disease, multiple drug candidates are currently undergoing clinical trials. Unfortunately, evaluating drug efficiency is a manual process that can only be done by a few experts. GDSC participants this year teamed up to create an AI-based solution to automate the process. To host the challenge, Capgemini partnered with AWS to provide an environment that would allow roughly 1,000 users to collaboratively train ML models on GPUs. At this scale, distributing capacities, securing the data, and managing costs are complex problems. This talk covers how they tackled these problems and what they’ve learned.

Our second talk will be about adopting mlops across multiple teams and products at GetYourgGuide, with Jean Machado and Mathieu Bastien.

Mathieu is a Director of Data Products at GetYourGuide, in charge of all the applied ML. Formerly was part of LinkedIn’s Data Science team and co-founded the open-source graph visualisation software Gephi.

Jean is a MLOps engineer and the Lead for the Machine Learning Platform team at GetYourGuide. In GetYourGuide he supports data scientists, analysts, and engineers with tooling and automation to build production-grade machine learning. Jean's has a major hobby with automation, and is excited about what machine learning and automation can do for society.

About their talk: Putting ML into production is hard, but what happens once you have dozens of ML models into production? How to keep creating product improvements while also caring about cross-teams ML systems good practices? In this talk we will present some of GetYourGuide’s learnings regarding ML systems operations with concrete examples of projects and strategies that worked for us.

The complete agenda is the following:

• 6:00 pm - Open the doors & Food
• 7:00 pm - Steffen's Talk
• 7:30 pm - Break
• 7:50 pm - Jeans & Mathieu's talk
• 8:20 pm - Lightning talks (5 min each)
• 8:35 pm - Socialize

If you are interested in giving a lightning talk, fill up the signup form on Meetup.com while confirming your attendance.
Lightning talk inspiration (thanks to MLOps.community Amsterdam):

• What are you working on?
• What do you (not) like about a particular MLOps tool?
• Why do you think XYZ is over/underrated?
• What are your takeaways from a book or paper you read?
• What concept or idea are you excited about?
• Did you discover any neat software packages recently?
• Do you have any great MLOps productivity tips?
• MLOops! -- Do you have any personal stories on ML gone wrong?