Machine learning at scale

We are back! It has been a while, and many things have happened in the world of data (and in the world in general). We are seeing widespread adoption of machine learning (ML) and data science technologies across all sectors, and the scale of both the models’ size and impact is rapidly growing. But while providing new opportunities, these technologies also bring a whole new level of complexity into the process of software development. This meetup will focus on the challenges inherent in scaling up ML model development and solutions to overcome them, and two experienced data science engineers will share their hands-on experience working with ML model development on a big scale. Afterward, there will be time to catch up with old connections or make new ones.

~ BASIC INFO ~

Date: April 21, 2022
Time: 18:00 - 21:30
Location: Mærsk, Esplanaden 50, 1098 København K

~ TALKS ~

How to iterate rapidly and reliably make a million daily predictions
Speaker: Søren Vind, Senior Engineering Manager leading the forecasting team at Mærsk
Søren will tell us about how they orchestrate forecasting in a way that supports solid and reliable operations, while also allowing fast experimentation loops.

Iterating on huge transformer models using Kubeflow pipelines
Speaker: Lasse Reedtz, Senior Data Scientist at Visma
Lasse will tell us about how they iterate on models that are too big to retrain often, and how they utilize Kubeflow Pipelines, neptune.ai and Vertex AI to run large-scale data processing and training pipelines with automatic caching.

~ AGENDA ~

18:00 - 18:30 Arrival, Food, and Drinks
18:30 - 19:00 Talk 1: How to iterate rapidly and reliably make a million daily predictions
19:10 - 19:40 Talk 2: Iterating on huge transformer models using KubeFlow pipelines
19:40 - 21:30 Meet like-minded people, make new friends and catch up with old ones.