Our client is a Seattle-based consulting company with expertise in using business and technology to empower its customers. Our client has large business companies as customers that would like to analyze their data flows and predict future trends.
Our client needed an application that could predict future trends.
Our team developed an application that predicts future trends so users can upload their data-train-sets (.csv files) to the system. These files are passed through a machine learning algorithm that learns this data and creates a data model. After that, the user can send new data sets (with conditions but without the results) and the data model to the ML service, which returns the results for the new conditions.
Benefits and Results
- Created an Angular 7.0 client-app that allows users to manage projects, upload data-sets, trigger processing of data-sets, and download the results.
- Improved the node.js API app: added ability to receive and save into AZURE Cosmos DB such entities as customers, taxonomies, datasets.
- Created a microservice that allows download/upload files to AZURE blob storage.
- Added a “supervisor” microservice that redirects KAFKA-messages from UI-topics to ML-topics.
- Added a job scheduling service that checks and removes old processing-requests from the storage over a specified period of time.
Docker, Kubernetes, Node.js, Angular 6, Kafka, Azure, Python, Visual Studio Online