Mourougan Peroumalle

Bpifrance is a public investment bank supporting businesses from their inception to IPO.

As a Tech Lead and Solution Architect, I played a pivotal role in modernizing and optimizing the company’s data infrastructure:

• Led the implementation of a data mesh and migration from legacy systems, achieving $200K annual savings and reducing data latency from T-4 days to T-1.

• Architected and deployed a real-time data lakehouse using Kafka and Iceberg for a speed layer solution.

• Managed a technical team of 4 developers and provided support to an operational team, ensuring smooth execution.

• As a Data Architect, ensured the coherence and evolution of the IT system by participating in the technical leadership committee for cross-functional projects.

Technologies: Python, AWS (S3, Glue, Lambda, StepFunction, Athena), Terraform, Kubernetes, Jenkins, Control-M, DataDog

Kayrros provides transparency to energy markets through satellite imagery, mobile phone geolocation data, ship geolocation, and more.

I led key data initiatives that drove efficiency and scalability:

• Initially joined the RD Data-Ops team to design and implement a data lake processing 300 million data points daily.

• Directed the satellite data acquisition team responsible for managing petabytes of satellite imagery, reducing incidents by 90% and ensuring SLAs were consistently met.

• Evolved the satellite acquisition platform to comply with STAC standards, transitioning from a data lake to a PaaS model for real-time satellite image acquisition and pre-processing, saving $200K annually in storage costs.

• Spearheaded the migration of on-premise infrastructure to AWS Cloud, improving scalability and performance.

Technologies: Python, Hadoop/Spark, Hbase, Elasticsearch, Airflow, Docker, Rancher, NodeJS, Golang, PostgreSQL, Kubernetes, Argo, AWS (S3, EC2), Rego

Adloox provides transparency in the advertising market by offering tools to detect fraud. I was instrumental in building the company’s data infrastructure:

• Worked alongside data scientists to create algorithms for fraud detection and implemented the company’s first Hadoop/Spark cluster, capable of handling hundreds of millions of requests per day for analysis.

• Developed scalable data processing pipelines to support real-time fraud detection to answers real time bids under 100ms.

Technologies: Python, Scala, LUA, Hadoop/Spark/Kerberos, MySQL, MongoDB, Ansible, Docker