Thibaud LE GALL

Servier Design and implentation of software pieces aiming at accelerating ML & RAG project development while providing out of the box reproducibility, logging, lineage (data, model, metrics) Architecture and development of CI/CD components and IaC to make thoose software pieces available accross teams and organisations within Servier Design of serving practice and deployment

Fnac darty Designing and implementing a dataplatform and operationnal databases for managing spare parts in relation with customer support and repair service offered by Fnac darty (from providing available references to technician, to serving data for analytics) Environment : Bigquery, CloudSQL, Cloudrun, Docker, Python, Gitlab, Terraform

Ag2r La mondiale Designed, evangelised and led a team of developper taking part in building the 1st dataplatform on s3ns (sovereign CGP offer) in order to migrate the previous dataplaform of the company Cloudera and build new use cases including data mesh, AI, analytics and more Environment : S3ns, GCP(resource manager, IAM, network, database and compute services), Terraform, Python, Scala (Cloudera), Java, Docker

Valéo Designed and implemented the 1st ML platform built on GCP with vertexai for advanced ML/AI model training bringing tools accross 40 teams for data versioning, experiment reproducibility, data and resources access management, infrastructure as code, CI/CD, analytics, coding practices and model deployement Environment : GCP(Vertexai, Bigquery, Cloudbuild, Cloudfunction, GCS, Artifact registry, IAM, Cloud source repository), Terraform, Kubeflow, Docker, Pytorch, Python, Bash

Renault - Logistics Assisting Renault in building new models for predicting demand of cars and parts. Implentation tools for logging allowing reproduciblity and comparability of experiments Environment : VertexAI, Jupyter, Bigquery

L'Oréal - R&I As a senior data engineer I work at the platform level on tools that helps data engineers to develop current and new projects more quickly and efficiently and facilitate the CI/CD process, as well as implementing tools and practices to ensure code quality. Environment: Terraform, GCP (Bigquery, Cloudbuild, GCS, Airflow), Github, docker, python(pre-commit, flake8)

Devfest Nantes, Devfest Strasbourg (Google Developper Group conferences) As a speaker, with my colleague Yulianna KHOROLICH, we presented a codelab around MLOps. Environment : GCP (VertexAI, Artifact Registry, GCS, IAM, Bigquery), Kubeflow (Kubernetes + tensorflow), python

Renault - Innovation and Logisitics As a senior Machine learning engineer I was in charge of providing a MLOps pipeline for deploying ML models in production and transforming datascientists code into production ready code, as well as providing guidelines, documentation, architecture recommendations Environment: GCP (VertexAI, Bigquery, Artifact Registry, IAM, Cloudbuild), Docker, Python (jupyter, sciPy, numPy, fastapi, kubeflow)

LVMH - PCIS As a data engineer I was in charge of building data pipelines and developing abstractions to be used by the team. As a data project manager I was in charge of supervising data projects for entities within LVMH like online marketing retargeting or kpi production Environment: GCP (Bigquery, GCS), Azure devOps, Dataiku, SQL

As a lead of data science, I work as a consultant for client as well as developping internal products and competences of people inside the team Management tasks : - Internal Skill development path - Recruitment process development - Management of up to 5 team members - Objectives definition - Annual meetings and reviews We also used our skill development program to build demonstrators that could be used as commercial arguments in convincing prospects Internal research projects included: - Recognition of books by their cover embedded on mobile(2021) - Video compression with AI models like AE and GAN (2020) - Live reskinning of objects on video streams (2020) - Implementation of A3C (reinforcement learning) for experimentation with internal apps (2019) - Facial Recognition for mobile, embedded model (2018) Environment: tensorflow, pytorch, pysyft, docker, android, AWS/GCP, javascript, github/gitlab

Wafa Assurance As lead data, I assisted Wafa insurance in setting up a digital factory and participated in building MVP of machine learning algorithms - Developped an MVP of an algorithm predicting the chance of a client to leave Wafa - Assisting Wafa in Setting up architecture and practice for developping AI algorithm - Managing the datat team and interaction with the business Environment: Python, Java, SQL, Docker

Ditto Bank At Ditto Bank, we worked on the mobile app and the back-end of the banking system to accompany Ditto bank in integrating and developping some new tools - Integrated a tool for identity recognition comparing face from a video to the photo of the id document and extracting informations from the identity document - Explored solution to build a service of automated credit agreement - Automated back-end tasks Environment: Python Jupyter, embedded, Infeedo

Airliquide Involvement in different projects for AirLiquide aiming to create value from data - Detection of machine's references on drawings of production line - Built an MVP for detecting churn on B2B market - Built data pipelines to feed new KPI in dashboarding tools Environnement : AWS, Python(Jupyter, Numpy, Pandas, Scipy, Scikitlearn, Tensorflow), Postgresql, PowerBI, Hadoop, Spark, MobatXterm

Egis In partnership with Egis, we developped a solution which enables Energy Managers to optimize the consumption of energy of the sites they are responsible for. - Development and integration of algorithms to predict consumption of sites - Development and integration of an algorithms and pipeline to qualify observed discrepencies of sites with their estimated baseline - Developped a recommendation system to push notifications to users Environment : Hadoop, Spark(Scala), R, Python, Docker, MongoDB, Jupyter, Jenkins Management of a team of 5 people