Machine Learning and MLOps for Home Value Adjustment
Redfin is a high-tech customer-first real estate brokerage firm for buying and selling homes. By combining its own full-service agents with modern technology they redefine real estate in the consumer’s favor. Homebuyers and sellers enjoy a full-service, technology-powered experience from Redfin real estate agents, while saving thousands in commissions.
Iron Mountain Incorporated (NYSE: IRM), founded in 1951, is the global leader for storage and information management services with 24,000 employees and more than 220,000 customers in 50 countries. Iron Mountain stores and protects billions of valued assets, including critical business information, highly sensitive data, and cultural and historical artifacts.
Redfin wanted to estimate the value of homes by including both image data and text into the assessment of a property. For example, when a user uploads an image of a property, a Machine Learning model can integrate a set of real estate features like the presence of a pool, fireplace, etc. to value the property.
SpringML developed Machine Learning pipelines, improved and automated Machine Learning practices on Google Cloud and Kubeflow. By structuring an MLOps practice with Redfin including CI/CD/CT infrastructure, the data science and IT teams collaborate and increase the pace of ML model development, training, and deployment to
By automating many ML models and deployment practices, the Redfin data science team is developing and deploying more ML models to the Redfin platform thus improving the customer experience. The Redfin team is looking at 3 other use cases to expand the use of their ML platform with Kubeflow.