Three years ago, I joined Ubisoft as a Data Engineer to work in a field I knew nothing about: transactional fraud in the gaming industry! 🎮

Time flew by quickly, and a few weeks ago, I said goodbye to the team as I have chosen to discover yet another field unknown to me… but I shall tell you more about that in a separate post 😉

Feature Store @ Ubisoft

What a thrill it was to be part of this amazing company! As data engineers, our work is often hidden behind the scenes, but I am so very proud of my achievements at Ubisoft 🚀

While the project revolved around detecting fraudsters, my challenge was broader: the goal was to design and implement a Feature Store that would centralize curated data to train Machine Learning algorithms.

Over the course of 2 years, by working hand-in-hand with an immensely talented team (Omni ❤), our Feature Store went from a few SQL scripts to a full-fledged framework, running in prod, where data is handled as code:

  • Features are self-explanatory thanks to detailed identity cards listing data sources, documentation and transformations. #PySpark #Python
  • Data materialization is optimized to handle upstream changes and fast querying. #Parquet #Iceberg
  • Our library not only produces training data, but also computes features on-demand for live inference. #AWS #DynamoDB
  • Observability is key: our monitoring system alerts us on key indicators for source and generated data! #Grafana #Prometheus

So much can be said about this project: it required me to delve into the meanders of e-commerce transactions, understand how our data scientists craft their features and their models, learn the subtleties of GDPR and personal data regulations, create roadmaps and communicate on updates. I learned a lot, and grew with the project even further by exploring and leading an initiative on #MLOps tools and processes!

Data Community @ Ubisoft

What I hadn’t planned on, was the chance to contribute to Ubisoft’s data community! 🤩

I co-organized Data Breakfasts 🥐 and Skill Sharing Sessions on topics covering the entire spectrum of data-related fields: NLP, data visualizations, Reinforcement Learning, data pipelines and formats, … a treasure trove of knowledge. I was also able to participate in the organization of the Data Science Days 2022 hosting speakers from Ubi locations across the world.

Thank you to everybody at Ubisoft for the amazing priviledge of working with and alongside you. It was truly an unforgettable experience and I leave with a heavy heart. But onwards we go… Looking forward to next chapter soon… stay tuned! ✨️