Reonomy is transforming the commercial real estate industry, connecting fragmented real estate data to help empower individuals, teams, and companies to unlock insights and discover new opportunities. To achieve their goals, they needed to overcome the challenge of unifying many disparate datasets with machine learning, which requires significant collaboration between data engineering and data science. With Databricks, they are able to unify data and AI, while simultaneously raising cross-functional collaboration and operating speeds to new heights.
To say that Reonomy started out with a complex mission would be an understatement. They knew that successfully connecting property, companies, and people would require building a framework integrated tightly enough that it would be a cinch for users to get answers to questions like, who owns this property? Who should I lend to? What’s the risk in working with these people?
“That sounds like a basic question, but oftentimes, commercial properties are owned by corporations, LLCs, and trusts,” explained Richard Sarkis, the company’s Founder and Executive Chairman. “Piercing through that corporate veil and building a knowledge graph and associating a company with a series of properties over time is massively valuable.”
In addition to dealing with disparate data sets, Reonomy also faced challenges associated with incredibly high volumes of data and the necessity of collaboration between data engineering and data science. Instead of the traditional data path — through the engineering pipeline, into a data lake and back out in model form — Reonomy’s models live within the data processing pipeline. Because those models ultimately define the shape of their data lake, cross-functional teamwork is an absolute requirement.
Databricks solves the extra-tall order for cross-functionality by providing Reonomy’s data teams and product management teams with an interactive environment to manage infrastructure and collaborate—from data ingestion and exploration to model training and deployment. Together, they’re now able to build models that are much more robust and ready for production thanks to this reduction in complexity: spinning up clusters is easy and cost-efficient, data engineers can quickly build data pipelines that scale for downstream analytics and ML, and data scientists can leverage interactive notebooks for faster iterations.
But perhaps most excitingly, using Databricks to unify data and build robust ML models has enabled the discovery of entirely new solutions for Reonomy’s customers.
“Databricks has been fabulously helpful with our ability to quickly change offerings and look for new product opportunities,” explained Maureen Teyssier, Reonomy’s Chief Data Scientist. “Our product managers actually work in the Databricks platform alongside data scientists and data engineers to identify alternate possibilities and new structures for new products, and this has greatly increased our ability to pivot to new spaces — a process that would otherwise take much longer.”
Velocity is a critical part of being a startup, whether it’s making sure that you’re iterating quickly, pivoting at a moment’s notice, or exploring all available opportunities for expansion. At the end of the day, Databricks powers the gamut of this standard for Reonomy, improving customer engagement and retention in ways that wouldn’t have been possible otherwise.
Prior to Databricks, the team at Reonomy were happy to deploy new ML models to production every 6 months. With a combination of Databricks and their highly talented data team, they were able to accelerate time-to-market of new models by at least 3x.
“Databricks has changed the way our data engineers, our data scientists, and our product managers all communicate,” added Maureen. “We can now take our models from virtually nothing to production in two months. It is truly the focal point for all of our product development.”
The landscape of commercial real estate has changed very dramatically as of late. With Databricks, Reonomy has the confidence and ability to react to market fluctuations and evolving customer needs. Through the velocity of their data organization and Databricks as their analytics backbone, they are now able to identify and act on new product opportunities — ultimately improving customer engagement levels and retention rates.
Databricks has changed the way our data engineers, our data scientists, and our product managers all communicate.”
– Maureen Teyssier, Chief Data Scientist, Reonomy