AutoML IL Relaunch - Let’s meet!

We are thrilled to announce the relaunch of the AutoML IL community with a brand new look, co-organizers, and an awesome meetup event!

Agenda:
18:00 - 18:30 We’ll enjoy some get-togethers, drinks, and food! 🍡🍹 🙂

18:30 - 19:00 MLOps to the rescue: How we 10x our DS Velocity with Model Manager, Naama Ziporin & Alex Ingberg @ Outbrain

19:00 - 19:30 Mastering the Ebb and Flow: Enhancing AutoML with Automatic Data Drift Detection, Noam Chiger @ Pecan

19:30 - 20:00 No Code Text Classifiers for Non-technical Users, Inbal Horev @ Gong

###########################################

Join us on June 21st for an engaging live meetup where you'll have the opportunity to listen to exceptional speakers, connect with fellow AutoML practitioners, and gain valuable insights into the latest advancements in the field.

RSVP to secure your spot!
Hope to see you there 🎉
Tuval St 40, 13th Floor, Ramat Gan

Abstracts:

MLOps to the rescue: How we 10x our DS Velocity with Model Manager
Naama Ziporin & Alex Ingberg @ Outbrain

Over the past 3 years, Outbrain’s Machine Learning systems were heavily reworked, helping us gain an exponential increase in our DS velocity.
In this talk, we will share one of the key enablers in this journey, a framework called Model Manager (MM) which allows data scientists to perform faster offline to online iterations. MM fully automated the pipeline to build, test, deploy, and monitor models for ab testing.
MM is a Python-based platform that seamlessly integrates Airflow with Outbrain's proprietary AutoML system to create end-to-end pipelines which support the DS work of Outbrain’s data scientists and ML engineers, running dozens of online models simultaneously.
In this talk, we'll share the motivation for building this framework, go through the general architecture, and showcase some of its capabilities in a demo.

Mastering the Ebb and Flow: Enhancing AutoML with Automatic Data Drift Detection
Noam Chiger @ Pecan

This presentation will spotlight the impact of data drift on machine learning models and the importance of automatic detection techniques for maintaining model accuracy. We'll explore core concepts, current automated detection strategies, and their real-world applications, offering attendees a concise, yet comprehensive overview of this critical aspect of data science. Ideal for data scientists and machine learning practitioners keen on safeguarding their model's performance over time.

No code text classifiers for non-technical users
Inbal Horev @ Gong

At Gong, we serve thousands of customers from various business domains. One of their main needs is to track nuanced concepts in their own data, such as customer objections or adoption of new messaging. To this end we developed Smart Trackers, empowering our users by enabling them to train their own tailor-made text classifiers without requiring any technical know-how.
In this talk we present our solution, describing the technical framework and highlighting the ML processes that we automated. We describe how we curate the dataset, and choose the model and its hyper-parameters, all while creating a simple, intuitive process. All the user needs to do is tell us about their concept and label a few examples.

###########################################

Don’t miss out ▶️ RSVP now