Thomas B.

The purpose of Idorsia is to discover, develop, and commercialize innovative medicines to help more patients. They want to transform the horizon of therapeutic options. In order to achieve this, Idorsia is developing into a leading biopharmaceutical company, with a strong scientific core.

Mission:

- Responsible for carrying on RWE Studies

- Responsible for designing a serverless MLOps platform to accelerate RWE insight generation

Technical Stack:

- Cloud: AWS SageMaker

- Back: Python, PySpark

- Automation, Support & Versioning: GitHub, GitHub Actions, Docker

The Direction de la Transformation Numérique (DTNUM), part of the Direction Générale des Finances Publiques (DGFiP) is responsible for the development of new data-driven strategies and workflows within the various departments of the DGFiP. Development of a state of the art AI platform, leveraging the latest MLOps tools and workflows to allow robust implementation of several use cases such as fiscal fraud detection or financially endangered company identification.

Mission:

- Responsible for the broader DTNUM AI & MLOps technical vision & strategy.

- Responsible for the design, architecture and production level delivery of a state of the art Machine Learning Platform leveraging the best Open Source technologies.

- Responsible for robust & RGPD compliant MLOps workflows

- Responsible for the success of various customer use cases on the platform

- Coaching of Data Engineers, Data Scientists and Machine Learning Engineers

Technical Stack:

- Cloud: Kubernetes, MinIO on top of OpenStack

- Back: Python

- Machine Learning: Tensorflow, Scikit-Learn, Jupyter, MLFlow, Kubeflow

- Automation, Support & Versioning : Terraform, GitLab, Kustomize, ArgoCD, OpenTelemetry, Docker

Development of agnostic cloud-based architectures allowing real-time communication for Computer Vision Deep Learning models to detect human movements during sport and re-education sessions. We collaborated with doctors and physicians specialised in sports as well as with LaBRI. We received the DeepTech label from BPI as well as several regional fundings. DeepMove was invited among 11 other highly promising startups to the Agoranov DeepTech Program.

Mission:

- Managed a team of 5 AI engineers & freelancers to develop, implement & deploy state of the art computer vision models as well as a Mobile SDK for easy integration of DeepMove technologies in customer applications.

- Designed & implemented serverless and microservice-based architectures on AWS leveraging the latest Infrastructure as Code practices to deploy AI models.

- Responsible for developing and putting high performance real-time & asynchronous video based AI services into production.

- Responsible for smooth technological integration with customers

Technical Stack:

- Cloud : Kubernetes, AWS: Lambda S3 SQS ECS EC2 Fargate ECR SageMaker Kinesis

- Front : Kotlin, Android, Swift for Mobile Apps and StreamLit for Web Apps

- Back : Python, Java

- Machine Learning : Tensorflow, PyTorch & Scikit-Learn

- Automation, Support & Versioning : Terraform, GitLab, Kustomize, ArgoCD, OpenTelemetry, Docker