In a recent panel discussion, Richard Zananiri, Director of EMEA Mid-market at Databricks, was joined by four globally operating, high-growth cloud-native companies. Each organization is at the forefront of utilizing the power of data, ML and AI to solve business-critical issues within their respective sectors. The company representatives were asked to explain how they are harnessing the power of data to overcome their biggest challenges, and how they are implementing Databricks technologies to reach scale and drive hyper-growth within their organizations.
Across all four companies, it was clear that by introducing the Databricks Lakehouse platform, data teams are able to manage and process data in a faster, more efficient and scalable way. This in turn allows them to achieve more with fewer resources.
Dodo Brands, a Russian tech-driven food service company that has experienced 40% growth over the last year, implemented Databricks to enable artificial intelligence (AI) and machine learning (ML) for predictive supply chain analytics and forecasting, which is instrumental to sustaining their competitive advantage. Their first project involved building a new infrastructure and system for supply chain planning, starting with short-term forecasting. This was then scaled up across the whole supply chain, ensuring that data was accurate, reliable and accessible for all their partners and employees. With around 600 pizzerias and 10,000+ employees, data can now be harnessed to provide all stakeholders with relevant information on how to increase revenues and manage their business effectively. The company’s next step will be to look beyond forecasting to future advanced analytics use cases, such as optimizing prices with ML algorithms. The aim is to complete projects up to 20 times faster, aligned with the rapid pace of the Dodo Brands business.
Fraudio is another fast-growing, disruptive business benefiting from the Databricks platform. The financial services company offers payment, merchant fraud detection and anti-money laundering solutions using their patented centralized AI technologies, and works with some of the biggest financial institutions in the world. Databricks’ solutions enable Fraudio to carry out data engineering, data science and model training and to manage business intelligence (BI) efficiently. Fraudio processes large amounts of transactions every day from different customers and third parties. Customer data schemas are translated to its internal data schema in real time, and all schemas are put together in one centralized place for the AI to be leveraged. The AI continuously learns from all the data received, producing a networking effect involving billions of transactions, enabling Fraudio to contextualize the data and produce extremely accurate scores that serve customers of all sizes to meet their real-world needs more rapidly.
With growth of more than 600% over the last two years, FollowAnalytics allows customers to build a mobile app, with the aim of growing mobile revenue streams and increasing marketing ROI. As its customer base has grown, FollowAnalytics has transitioned to Databricks to manage each customer experience individually with separate flows. Over the next 18 months, the company is moving entirely to the Databricks platform to increase flexibility and business impact for customers, as well as managing internal resources more efficiently. With the large volume of data from some retail and e-commerce customers, FollowAnalytics has developed specific analytics solutions customized to each customer, and is starting to implement AI and ML models that allow automatic segmentation. Modeling is completed on a per client basis, so that customer data is not mixed – ensuring compliance with data protection and privacy requirements. Another key advantage of Databricks for FollowAnalytics is stability. Different customers have subtle differences in their applications, so with A/B testing, they can measure which solution serves the customer best, increases revenue and maximizes ROI. With several million people using the applications daily, the FollowAnalytics team confirms that this would not be possible without Databricks.
S4M is a rapidly-growing business that delivers advertising designed to drive customers to stores, dealerships and restaurants. Its goal is to increase value from campaign spend, optimizing and reporting on business KPIs such as store visits and sales. More than 1,000 brands use the S4M Fusio platform to drive customers to physical locations. The company transferred to the Databricks open source Delta Lake platform, which has increased performance and provided full integration with open source services such as MLFlow. Its data teams have transitioned from batch jobs to streaming, which allows them to segregate and create a layered architecture. This means they now operate with more agility and flexibility when feeding ML jobs. Thanks to Databricks, S4M can solve an array of critical problems such as qualifying bids, how much should be paid for performing campaigns and geolocalisation fraud issues.
The S4M data team uses MLFlow daily on production jobs or using notebooks. MLFlow is employed to track every model, run and record. The data can be versioned by using just a few more lines of code, making operation much simpler, and iteration much faster. All actions are completed ‘under the hood’, saving time and effort: the tracking of the runs, the parameters and the metrics, and models can be automatically exported to S3. Furthermore, Databricks makes it easy for S4M teams to change cluster configurations, to bootstrap jobs and to go on production every couple of hours. It has also helped the teams embrace a layered architecture, which can be leveraged much more easily using Delta Lake atop their data lake. In summary, Databricks has made S4M’s data operations more cost-effective and efficient, allowing teams to accomplish more with reduced resources.
In conclusion, all the organizations on our panel – although operating in vastly different industries and customer sectors – share similar goals in terms of data management: driving efficiencies across their teams and moving forward with agility and speed.
We would like to thank our guests and moderator for their participation and insights. Watch the full discussion through the link below.
Jose Carlos Joaquim – CTO, FollowAnalytics
Joao Moura – CEO, Fraudio
Michael Colson – VP Platform & Data, S4M
Clement Carreau – Data Engineer, S4M
Vladislav Mandryka – Global Supply Chain Director, Dodo Brands
Andrey Filipev – Chief Data Officer, Dodo Brands