Munich Data Science #2: Data Efficiency in action

We are so excited to meet again after such a long break!

The topic of our next event is Data Efficiency: how to succeed on every scale of data.

This time, we have two great talks for you: Vsevolod will talk about ML pipelines on a petabytes scale, and Ivan will show us how to win with small datasets over big ones.

We want to thank JetBrains for providing us with a cozy venue!

=== Agenda ===
18:40 - 18:45 Welcoming and Intro
18:45 - 19:00 Host talk: Data Science at JetBrains
19:00 - 19:35 Talk 1: How to win with small datasets over large ones.
19:35 - 19:55 Networking
19:55 - 20:45 Talk 2: Storing and using data on 3 petabytes scale
20:45 - 21:30 Networking

=== Talks ===
Vsevolod's talk description:
3 billion user profiles and 3 petabytes of data. How to store it, use it and make ML pipelines on such scale.

Including a bird-high overview of our:
- servers and storage systems: hardware and software
- data processing and retrieval
- The bottleneck of machine learning and the trade-off of quantity/quality of the data
- Impressed by numbers? We are actually data poor, and I will show why

Vsevolod Bio:
Seva spent last 5 years on solving various tasks in the domains of reinforcement learning, bidding and game theory.
Now here, at Bidease, all of this is combined with the real Big Data amounts, clusters and low latency processing, making the challenge even more tough and interesting.

Ivan talk description:
How to win with small datasets over large ones.

Infineon is building smart sensor solutions for IoT devices such as smart lamps with hand gesture controls. As part of the radar software R&D team, I would like to tell a story of how we collected recordings from hundreds of people and then find a way how to get significantly better performance with the data from just the two persons.
I will also show how we develop ML algorithms that should run with low response time on an MCU with 120kB RAM.

Ivan Bio:
Ivan is a Senior Machine Learning Engineer at Infineon Technologies AG.
His work is to combine ML models with conventional physics-based signal processing to run algorithms on edge devices. Apart from that, his experience lies in the area of Computer Vision, Medical Imaging, and ML for mobile devices.

See you soon!