Capital One at KDD 2025: advancing AI & data science

The KDD 2025 (Knowledge Discovery and Data Mining) conference is just around the corner, running from August 3 to August 7 in Toronto, Canada. As one of the world’s premier data mining conferences, KDD is a valuable opportunity to explore the latest AI research, machine learning applications and innovations in data science.

Capital One is proud to participate in this year’s conference, sharing insights from our ongoing work in AI and tech innovation. From advancing proprietary AI tools to applying cutting-edge models to complex financial data, our researchers and engineers are helping shape the future of AI in financial services.

We have several exciting engagements planned throughout the conference, highlighting how Capital One technologists are contributing to real-world progress in machine learning, generative AI and scalable systems built on a modern tech stack.

KDD 2025 workshop: machine learning in finance

• Details: Monday, August 4th from 8:00 a.m. to 12:00 p.m. EDT
• Location: Conference Room 603

The 8th annual Machine Learning in Finance Workshop provides a platform for researchers and practitioners to explore the evolving landscape of AI in the financial sector. Senthil Kumar, Head of Emerging Research at Capital One, serves as one of the workshop organizers.

The financial industry generates a vast array of data, from structured tabular datasets to intricate time series and dynamic click streams. This workshop aims to bridge the gap between traditional models for tabular data analysis and the emerging potential of Generative AI to transform how the industry can approach time series, click streams and other unstructured data forms. Discussions will focus on uncovering new methodologies and techniques to enhance predictive analytics, fraud detection and customer insights.

The workshop will feature a robust program, including invited talks from leading experts across academia and industry, as well as presentations of accepted papers. Topics covered range from understanding and predicting customer behavior, detecting anomalies and developing agentic AI for personalized experiences, to critical areas like model safety, explainability, governance, fairness, and privacy preservation in financial contexts. By bringing together diverse perspectives and encouraging short papers from industry practitioners, the workshop will foster collaboration and inspire new research directions that address real-world challenges in finance. 

KDD Finance Day: financial sequence modeling

• Details: Monday, August 4th (additional information on KDD Finance Day Site)

As a part of KDD Finance Day, Igor Melnyk, Distinguished Applied Researcher at Capital One, will present "Modeling Sequence Data in Financial Services." His talk will highlight the inherent sequential nature of financial data, which includes transactions, payments and digital activity, and the challenges this presents, such as data sparsity, varying temporal granularity and diverse schemas. The presentation also reviews various machine learning approaches for both non-sequential and sequential tabular data. His talk will delve into industry-leading approaches for modeling sequential financial data, such as digital activity and transaction histories, to enhance customer experiences. He will cover key factors like joint versus individual stream modeling, data representation, time modeling, attribute ordering and tokenization strategies. Attendees will also learn about future research directions, including Generative AI, Graph Neural Networks and Reinforcement Learning in financial contexts.

Invited talk: from features to sequences—transforming financial modeling at Capital One

• Details: Wednesday, August 6th from 12:00 p.m. to 12:30 p.m. EDT
• Location: Hall E, Main Stage 

Our data scientists Jan Amtrup and Victoria Martins will deliver their talk, "From Features to Sequences—Transforming Financial Modeling at Capital One," as a seated lunch conversation. They'll discuss the evolution of data science and machine learning at Capital One, with a focus on how modern sequence modeling techniques are being used in fraud modeling.

The talk will explore the role of transformer architectures in addressing challenges faced by traditional tree-based models, particularly when dealing with sequential data in payments. Jan and Victoria will share how sequence models have been applied to fraud detection, demonstrating improvements over traditional approaches by capturing complex temporal dependencies in financial data.

Connect with Capital One at KDD 2025!

Capital One will have a booth at KDD 2025. Our team members will be on the ground participating in discussions and are available to share more about our work in data science and machine learning. We invite you to stop by and connect with us to learn about the research and applications our teams are developing.

• Visit our booth: 7
• Explore our research: Dive deep into our latest advancements in AI and machine learning.
• Discover career opportunities: Learn about exciting applied research career paths at Capital One for researchers and engineers passionate about AI and join our world-class team.
• Learn about our student and grad internships: Put your knowledge and skills to work in our 10-week to two year graduate programs innovating new products and creatively solving the problems that impact our customers and our business.
• Engage with our team: Meet our researchers and AI experts, explore how we’re shaping financial services with patented AI and discuss what’s next for AI in finance.