Best practices for building machine learning platforms on the cloud
This article is part of a VB Lab Insights series paid for by Capital One.
Most people are familiar with the major technology platforms like iOS, Windows, and AWS. Platforms, in essence, are a group of technologies that serve as a foundation from which to build, contribute, experiment, and evolve other applications. They enable many of today's most advanced technological capabilities and cutting-edge customer experiences.
In order to keep pace with the scale and complexity of technological capabilities brought by big data, AI and machine learning (ML), many companies are developing their own sophisticated internal platforms. In fact, Gartner predicts that by 2025, cloud-native platforms will power more than 95% of new digital initiatives, up from less than 40% in 2021.
In my experience, enterprise technology platforms have been transformational: they allow cross-functional teams to test, launch, and learn at a rapid pace; reduce duplication and standardize capabilities; and provide consistent and integrated experiences. In short, they help turn technology into a competitive advantage.
Increasingly, organizations are becoming better at delivering best-in-class customer experiences by leveraging cloud-native platforms such as Kubernetes, capable of running heavy AI workloads and ML. Capital One's decision to become the first US financial institution to leverage the cloud and our ability to revamp our data environment has been integral to expanding the capabilities of our cloud-based platform. With this solid foundation, we are better able to leverage big data to build new ML capabilities on top of our enterprise platforms to accelerate, improve, and deliver new customer experiences. more meaningful.
Much of our work in this area is already yielding significant results for the company and for our customers. For example, our Fraud Decisions Platform was designed from the ground up to make complex decisions in real time. By leveraging massive amounts of data and enabling model updates within days (instead of months), the platform helps protect millions of customers from card fraud and can be used by a variety of company stakeholders.
Based on my experience leading teams delivering enterprise technology platforms, I learned some important lessons and best practices:
It all starts with the team: Build a cross-functional team of the best people, even if it slows you down at first. A bigger team is not always better! At a minimum, the team should have product managers, engineers, and designers. Staff these functions with people who truly understand platform users. For example, if you are building a platform that will be used primarily by data scientists, hire a product manager who was a data scientist, or put a data scientist on your leadership team. If the team is made up of people from multiple organizations, make sure you have common goals. Work backwards from a well-defined end state: Before you start creating, take the time to align yourself with the end state architecture and your plan to navigate your way to that destination. Make sure your architecture is designed for self-service and contribution from the start. Better yet, design the platform assuming you will extend it to users outside of your immediate organization or industry. Suppose over time you want to swap components as technology evolves. Estimate how long you think it will take, then double it: It's important to take the time to think about all the abilities you need to develop from the start, and then create a level of effort the size of a t-shirt for each component. Once your tech teams have paired this with velocity to estimate how long it will take to build each feature, add a 50% buffer. In my experience, this estimate ends up being surprisingly accurate. Focus on business results: Building successful platforms can be time consuming. It is important to sequence work so that business value can be achieved along the way. This motivates the team, builds credibility and creates a virtuous circle. Be radically transparent and over-communicate: Generously share decisions, progress, and roadmaps with stakeholders. In addition to articulating what you're working on, also articulate what you're not doing.
This article is part of a VB Lab Insights series paid for by Capital One.
Most people are familiar with the major technology platforms like iOS, Windows, and AWS. Platforms, in essence, are a group of technologies that serve as a foundation from which to build, contribute, experiment, and evolve other applications. They enable many of today's most advanced technological capabilities and cutting-edge customer experiences.
In order to keep pace with the scale and complexity of technological capabilities brought by big data, AI and machine learning (ML), many companies are developing their own sophisticated internal platforms. In fact, Gartner predicts that by 2025, cloud-native platforms will power more than 95% of new digital initiatives, up from less than 40% in 2021.
In my experience, enterprise technology platforms have been transformational: they allow cross-functional teams to test, launch, and learn at a rapid pace; reduce duplication and standardize capabilities; and provide consistent and integrated experiences. In short, they help turn technology into a competitive advantage.
Increasingly, organizations are becoming better at delivering best-in-class customer experiences by leveraging cloud-native platforms such as Kubernetes, capable of running heavy AI workloads and ML. Capital One's decision to become the first US financial institution to leverage the cloud and our ability to revamp our data environment has been integral to expanding the capabilities of our cloud-based platform. With this solid foundation, we are better able to leverage big data to build new ML capabilities on top of our enterprise platforms to accelerate, improve, and deliver new customer experiences. more meaningful.
Much of our work in this area is already yielding significant results for the company and for our customers. For example, our Fraud Decisions Platform was designed from the ground up to make complex decisions in real time. By leveraging massive amounts of data and enabling model updates within days (instead of months), the platform helps protect millions of customers from card fraud and can be used by a variety of company stakeholders.
Based on my experience leading teams delivering enterprise technology platforms, I learned some important lessons and best practices:
It all starts with the team: Build a cross-functional team of the best people, even if it slows you down at first. A bigger team is not always better! At a minimum, the team should have product managers, engineers, and designers. Staff these functions with people who truly understand platform users. For example, if you are building a platform that will be used primarily by data scientists, hire a product manager who was a data scientist, or put a data scientist on your leadership team. If the team is made up of people from multiple organizations, make sure you have common goals. Work backwards from a well-defined end state: Before you start creating, take the time to align yourself with the end state architecture and your plan to navigate your way to that destination. Make sure your architecture is designed for self-service and contribution from the start. Better yet, design the platform assuming you will extend it to users outside of your immediate organization or industry. Suppose over time you want to swap components as technology evolves. Estimate how long you think it will take, then double it: It's important to take the time to think about all the abilities you need to develop from the start, and then create a level of effort the size of a t-shirt for each component. Once your tech teams have paired this with velocity to estimate how long it will take to build each feature, add a 50% buffer. In my experience, this estimate ends up being surprisingly accurate. Focus on business results: Building successful platforms can be time consuming. It is important to sequence work so that business value can be achieved along the way. This motivates the team, builds credibility and creates a virtuous circle. Be radically transparent and over-communicate: Generously share decisions, progress, and roadmaps with stakeholders. In addition to articulating what you're working on, also articulate what you're not doing.What's Your Reaction?
