As one of the first pillars of Callyope, I lead the data infrastructure part, designing and building robust systems on the cloud following best practices, using mainly Python, Terraform and AWS.• Designed and implemented a secure daily backup system to transfer files and database dumps from AWS to GCP. • Built a data warehouse and an end-to-end ELT process using Python and DBT, enabling researchers to create robust training corpus datasets and the operational team to monitor clinical trials effectively. • Developed a serverless architecture for computing tasks and a warehouse querying API, slashing infrastructure costs by over 90% while implementing near-instant auto-scaling and ensuring cloud-backed uptime. • Implemented company-wide CI pipelines for Python/Terraform linting and introduced a Docker workflow, enhancing code readability and reducing deployment issues by over 80%, significantly cutting troubleshooting time from days to hours.

In charge of assisting a dozen Data Scientists from nothing to production.• Built a custom job orchestrator tailored to data scientists’ needs to distribute computing with MapReduce using Python, MongoDB, EC2 and Docker, significantly speeding up computations from hours to minutes and increasing cost efficiency by over 80%, from thousands to hundreds of euros. • Implemented a serverless API to interact with the job orchestrator, reducing API infrastructure cost by over 60% while ensuring near-instant auto-scaling, cloud-backed uptime, and secure authentication. • Developed several monitoring solutions, including periodic checks on team services, which reduced bug detection time by over 80% (from hours or days to minutes), and improved access to machines Docker logs, cutting troubleshooting time by half. • Contributed to the development of CI/CD pipelines, incorporating linters for best practices and automating deployment of the latest versions of infrastructure and applications from development to production, accelerating development time by 30% and reducing human errors by over two third.

Mise en place d’outils de traitement de données au sein de Sensome, une start-up issue de l’Ecole Polytechnique spécialisée dans la création de dispositifs médicaux connectés innovants.Utilisation et comparaison de plusieurs bases de données sur le Cloud et choix d’une architecture adaptée aux besoins de l’entreprise.Analyse de données et recherche de corrélations dans les rapports fournis par les médecins dans le but d’améliorer le produit (un guide pour aider les médecins à reconnaitre le type de caillot formé lors d’un AVC). Utilisation d’algorithmes de Machine Learning et analyse d’images par segmentation, le tout en Python.Outils : Python, Google Cloud Platform (GCP), Data Store, BigQuery