Context and problematic
Our client wants to possess a good vision of the current and future needs of his clients. Therefore, it was all the more important to aggregate a very large amount of data on each of the consumers. After a successful POC, we worked on the development of a platform for daily reporting.
Goals
Prove that the project is feasible with a first POC of a few days
Ingestion and pre-cleaning of data providers while industrializing the analysis and prediction models developed
Allow end users to visualize current and future needs
Our intervention
2 Data Scientist, 1 Data Engineer, 1 Data Architect, 1 Full-Stack developer
- Definition of the scope and launch of a pilot project to validate the feasibility in 8 weeks
- Implementation of the project architecture on AWS
- Integration of machine learning algorithms
- Development of analysis and prediction algorithms
- Development of the visualization application in reactJS and VueJS
Résults
Platform is functional since December 2019 for the business across all France entities.
In the industrialization phase for the customer at the group level.
Even till date, we are working on the analysis of new perimeters such as its e-reputation.
Technical environment
Python – Go – Typescript – Sklearn – reactJS – VueJS – NodeJS
AWS S3 – VPC – IAM – Cognito – ECS/ECR – Lambda – Glue – RDS – Fargate
Batch – WAF – API Gateway – DynamoDB – ElasticCache – Route53 – EventBridge – SageMaker Elastic Search – Kibana
Docker – Kubernetes – Terraform