Learning-To-Rank Framework - how to train an army of models?

At OLX, we faced a significant challenge: the need to experiment with rankings across many countries and categories, totaling over 100. This situation became a bottleneck, hindering our Data Scientists from progressing in their work. Given our limited human resources, it was imperative to seek more automated solutions. We implemented LTR Framework, allowing our Data Scientists to adjust configurations and select precise features and targets for optimization. The models are stored in a Model Store, where they can be deployed for A/B ranking tests. This project addressed the bottleneck issue related to experimenting with numerous models, enabling our business to test new ranking models effortlessly. The ability to conduct rapid and numerous tests can accelerate the growth of successful events on the platform and enhance our understanding of user relevance preferences. Furthermore, LTR Framework is easily transferable to other businesses in need of experimenting with LTR/ranking models.

Marcin Gumkowski

OLX Group

I’m a Senior Machine Learning Engineer at OLX Group with various experiences ranging from software development, data processing to training and serving machine learning models. During my last few years, my focus has been on building and improving Search products with the usage of Machine Learning and Cloud Computing.

Catarina Gonçalves

OLX Group

As a Junior Data Scientist at OLX, I focus on solving ranking problems, also leveraging my background in ML to drive insights and innovation. Passionate about Data Science, I am dedicated to expanding my knowledge and expertise in this fast-evolving field.