Industry Retail
Region North America
Solution End-to-End Data Platform Enhancement
Context
Over the past decade, data analytics has emerged as a game changer, and a data platform that delivers quick and informed insights has evolved into an essential asset that retail firms must incorporate into their strategies to maintain competitiveness. Retailers that began building an end-to-end AI-powered data analytics platform are realizing the immense effort and expertise required to make it fail-proof. As a result, retailers are seeking data engineering and machine learning operations (MLOps) experts to develop a robust solution that elevates their MLOps maturity and helps achieve their end goals.
Problem statement
Our client, a supermarket chain, wanted a cost-effective solution for boosting their data engineering maturity in the Azure ecosystem. Their Operations team wanted to develop a modular, scalable, and secure framework in Airflow with Databricks that would embed the best engineering and MLOps practices to productionize and optimize data workflows while reducing costs and overall runtime. Additionally, they intended to automate the repetitive tasks and utilize model monitoring to manage the platform.
Impact
- Reduced runtime of Merchandizing MLOps initiative by 65%
- Decreased cost of Operations department by 80%
- Reduced Personalization runtime by 48%
- Saved Databricks unit (DBU) hours by nearly 340 hours per month and achieved annual savings of over $180k for the company