Hammad Squalli-Houssaini

Digital solutions developed by Wüest Partner provide clients clear overviews and insights in the real estate sector. These solutions are based on a variety of data which real estate experts rely on for precise estimations. As a senior data engineer, I contributed in implementing scrapers and parsers in Python that collect useful data; Keep MongoDB databases updated; Implementation of geospatial algorithms that divides France to rasters of 25x25m resolution and predicts evaluations of each grid cell given defined criteria. Plannification of diferrent steps of the project. Management of a team of data scientists and data engineers. Implementation of statistic methods based on raster files and geospatial data. Manipulation of raster files. Finally, use of Google Cloud Platform for better performances.

Orange owns several datacenters in France which are mutually used by thousands of projects. Managing the infrastructure resources of CPU, RAM and disk storage in such a centralized environment requires data science expertise, hence my role at the team GRIN of the infrastructure direction. As a data scientist, I worked in a Spark and Hadoop environment on prediction of infrastructure resources consumption by projects, as well as correlation between their preemption and ground truth consumption. Part of my work was to provide clear communication about current state of used and available resources. I managed to do so using Kibana dashboards. Finally, I developed workflows using Python, Ansible and Gitlab to make it easy for infrastructure experts to deal with some of their regular tasks.

In Thales Reseach & Technology, I worked in collaboration with Thales Alinea Space on test bench data of satellite components. These data come regularly in high volume and their interpretation is often not evident by test bench experts. That's why I used statistics and machines learning techniques to help taking full advantage of these data. I did so using Python for algorithm implementations, and Kibana for dashboard generation. I also participated during this project on data mining on tweets of control agents. These tweets come in important flows of data and need to be processed in real time in a a scalable solution. I managed to do so with the Elasticsearch stack.