Home 5 Projects 5 Machine Learning Platform @ BNP Paribas, 2022-2024

Tech stack:

Openshift

Openshift

Helm

Helm

Docker

Docker

MLFlow

MLFlow

JupyterHub

JupyterHub

JupyterLab

JupyterLab

Hue

Hue

YAML

YAML

Click to Expand section and show all tech used for this project

Story

At the beginning of 2022, an organizational change at my workplace took place – Agile@Scale. It has been a new way of working for multiple departments – aimed to be faster and more effective than before. As a byproduct of these results, I got reassigned to a newly-formed Squad – whose audacious goal has been to create an innovative and comprehensive Machine Learning Platform for Machine Learning Engineers, Data Scientists/Engineers/Analysts employed at Bank.

 

Solution

The described BNP Machine Learning Platform is architecturally consisting of multiple parts, powered by and fully-deployed at an on-premise cloud environment. Functionally, it is a mix of open-source frameworks, and a big data cluster, together with custom-tailored implementation of the SOTA market data science and machine learning solutions.

 

Sample Architecture

 

 

Lesson Learned

The scale and scope of this project is indeed audacious – thanks to this factor, I am not only able to learn blazingly-fast, but also take up, design and discuss new challenges every work day.

        Statistics

        • Not available – the whole project has been a private property of BNP Paribas Bank Polska S.A.