From click to predict and back: Data Science pipelines at Odnoklassniki

День 1 /  / Зал 1  /  RU / For practicing engineers

Machine Learning is fun, but in order to make it work in industry a lot of boring stuff needs to be done. In our talk we will consider all the technologies, algorithms and methods needed to make your ML shine like a diamond in a beautiful setting.

As an example, we will consider a single, but complex task – news feed personalization. Without diving into ML details, we will talk about real-time and batched data collection, ETL, and processing needed to get a model.

But getting the model is not enough, thus we will also discuss the way how to get model predictions in a complex high-load distributed environment and use them for decision making.

The talk covers different data processing and storage technologies from Hadoop ecosystem and more. If you are doing ML not just for fun, but also for profit, you might get some useful information from the talk.

Download presentation