Advancing state-of-the-art science and AI in financial services (AIM263)
Scaling AI at Capital One: A Journey of Transformation
Introduction
James Montgomery, Director of Data Science at Capital One, discusses the company's approach to AI at scale and its integration of cutting-edge AI research in the finance industry.
Provides a brief overview of his background and the transformation he has witnessed at Capital One over the past 10 years.
Capital One and Machine Learning
Capital One is a relatively young bank (about 30 years old) that has evolved from a credit card company to a large retail and commercial bank with over 100 million customers.
Leverages machine learning in various areas:
Adversarial use cases: Fraud detection, cybersecurity, anti-money laundering
Marketing and customer acquisition
Optimizing customer and employee experiences
Lessons Learned: The Human Component
Enabling cross-functional partnerships is crucial for scaling AI responsibly:
Diverse research teams with data scientists, engineers, business/product partners, and applied researchers
Collaboration with lines of business, risk/compliance teams, and other internal stakeholders
Early and frequent engagement to align on strategic priorities and navigate challenges
Partnerships beyond Capital One's walls:
Collaborations with top-tier academic institutions
Partnerships with technology providers (e.g., AWS) to leverage expertise and platforms
Lessons Learned: Modernizing the Tech Stack
Transition to the cloud and standardization of the data ecosystem:
Increased flexibility, reliability, and accessibility of data
Enabled the use of open-source software and modern tools
Significant investments in building a talent pool of 14,000+ technologists
Strategic initiative to deploy AI at scale throughout the organization:
Collaboration between data scientists and engineers to address challenges
Adoption of tools like Kubeflow to improve efficiency and scalability
Recent Initiatives and Challenges
Personalized and proactive customer communication:
Leveraging heterogeneous, dynamic, and tabular data
Addressing challenges in model deployment and integration with customer-facing systems
Improving employee efficiency through generative AI-powered agent servicing:
Integrating language models, vector databases, and real-time inference
Ensuring robust evaluation, testing, and governance to mitigate failure modes
Conclusion
The journey of scaling AI at Capital One has been challenging but rewarding, with a continued focus on responsible innovation and cross-functional collaboration.
The company's investments in modernizing its tech stack and building a diverse talent pool have enabled it to push the boundaries of AI research and application in the finance industry.
These cookies are used to collect information about how you interact with this website and allow us to remember you. We use this information to improve and customize your browsing experience, as well as for analytics.
If you decline, your information won’t be tracked when you visit this website. A single cookie will be used in your browser to remember your preference.