Raphaël B. Email & Phone Number

Lyon, Fr

France

Paris, Île-De-France, France

I manage the development of the new data processing platform within a team of 4 developers.The new platform is scalable to perform data analysis, enrichment and graph computation of multiples projects in parallel, each one containing from 1M to 40M documents to be processed.With the new architecture, the average processing time decreased from 14 hours to 3 hours, strongly reducing the number of failures during the processing workflow and allowing the end users to be more comfortable in their work of data analyze, allowing them to process more data by providing fully processed data in a reasonable delivery time.The new architecture mixes streaming and batch processing to provide a very fast orchestration of each analysis step.## Streaming architecture* Microservices following the streaming enrichment pattern using Kafka as data source and output* Subset of data flowing through each microservice can be easily invalidated without performing manual operation nor topic cleaning* 0 data loss with a at least once consuming strategy* Autoscaling managed by Kubernetes## batch analysis architecture* Highly scalable data platform using AWS EMR autoscaling* Predictable workflows with failure recovery using airflow pipelines* Handle multiple datasources like Amazon S3, RDS/PostgreSQL, Elasticsearch, Kafka* Idempotent jobsServices are monitored using Prometheus, Grafana and Kibana

Paris, Île-De-France, France

Provide infrastructure support, guidelines and best practices to the data scientists teams in the building of their data platforms.As a result of their ambition to become an innovative data driven company, Pernod Ricard created the Data Center of Excellence in 2020 to increase their capabilities of developing well suited cloud data platforms.As a Data architect, I was in charge of providing the good architecture and tooling to allow data science teams to reduce their delivery time by speeding up their development and model training phase, and to design a robust architecture, to move from an advanced MVP to a scalable production grade product.## Tech stackThe new data platform is built using microsoft's Azure cloud:* Azure Functions / Durable Functions for asynchronous services* Azure Machine Learning for model training and heavy resources computing data science applications * ADLS storage for raw and mid-data ; refined data are stored in Snowflake* POCS: * Programmatically schedule long running data sciences jobs on Azure ML env * Move from single threaded application to distributed computing Databricks/Spark * Using Azure pub/sub services to increase communication capabilities between data science processes and scheduling applications * Evaluation on Azure Synapse Analytics as a new data integration and Query as a service platform* Applications are developed exclusively using Python

Lyon, Auvergne-Rhône-Alpes, France

# Core-data Engineering Team Leader (2020):Within a 4 developers team, I've been a part of the design and managed the new data integration platform. We've collected raw data, stored it in an efficient way and exposed some enriched data as a self service analytics catalog or in more structured views like Datamarts.I worked closely with a partner to the design and development of a machine learning data quality alerting system to avoid data skew and avoid making available unexpected results to our clients.# Senior Data Engineer A/B testing team (2019):- Wrote the new A/B testing application to be cloud compliant- Designed a generic KPI computing algorithm- Applied and shared functional programming best practices

Région De Lyon, France

# Data Centric Application Technical Lead (LIZEO / 2017-2019):Within a team of 4 developers, I helped designing a generic real time data matching engine based on Kafka, with a goal to handle half a billion events each day.I took the lead of the engineering process for the new steps to be added, and for the design of the Backend, based on the Ports & Adapters architecture. This design strongly helped us to plug each new component without making the service unavailable, and made the development comfortable by making each technical component easily testable.# Software Lead Developer (2016-2017): - Designed and developed a building configurator (Lafarge Holcim / 2017) - PHP 7, React/redux/Javascript - Turned a legacy Drupal application to a microservices system (Attestation Legale / 2016) - Scala/Finatra - Developed some Single Pages Applications (AXA / 2016) - PHP 7, React/redux/Javascript

Région De Lyon , France

# Software Lead Developer (April / 2015):Within a team of 3 developers, I led the development of a backend for frontend responsible of exposing customers healthcare data.The services aggregated data from several backend like SOAP and REST API, stored in MongoDB as a temporal cache datastore.# Software Developer (Auvergne Rhône-Alpes / 2015):I developed some features for a website dedicated to expose travel trips and good addresses in Auvergne Rhône-Alpes country.# Software Developer (LIZEO / 2012 - 2014):I joined a team of 3 developers to develop and maintain a tyres comparison website.I was responsible of the development of the new CMS component, with the goal of designing a generic blog post platform in more than 15 languages.

Lyon

I developed some e-commerce websites and web applications (Facebook, Spotify)

Région De Rhone-Alpes, France

I worked on a web application dedicated to customizing and batch emailing promotional and corporate documents.

Dut, Informatique

Année spéciale : programme du DUT condensé sur une année

Maîtrise, Science Politique

Mention sociologie des organisations

Licence, Sociologie