Capital One’s Applied AI research

At Capital One, we are driven by our mission to change banking for good by helping improve the financial lives of millions of customers. As part of this journey, we’re accelerating the adoption of state-of-the-art AI research into our business to build new customer experiences and drive transformational business outcomes.

Our AI Foundations team, which includes our Applied Researchers, is at the center of bringing to life our vision for AI at Capital One by working on some of the most interesting and forward-leaning challenges at the intersection of AI and finance. Together with our partners inside and outside Capital One, we're fueling the frontiers of state-of-the-art AI research in finance to help make banking simpler and safer for over 100 million customers.

Our Applied AI research focus areas

Capital One’s Applied research team is advancing the frontiers of large language models (LLMs) and agentic AI. Our work explores the full lifecycle of LLM development—training, customization and alignment—with emphasis on reasoning, synthetic data generation, continual-learning and multi-agentic based complex reasoning. We aim to build models and AI systems that can learn, reason and act autonomously to solve complex real-world problems.

Foundation models and architecture

Focuses on the development and fundamental characteristics of large-scale, foundation models across various data modalities. Includes model types like Language, Graphs and Time Series, pre-training and fine tuning strategies, and how to integrate Knowledge Systems and Multimodal Knowledge representation to enhance factual understanding.

LLMs and agentic AI

Dedicated to building next-generation AI systems that act and reason independently and collaboratively. Includes single- and multi-agentic systems, advanced complex reasoning techniques, resource management models like Test Time Compute and Mixture of Experts, and the use of Reinforcement Learning, including RLHF and World Models, to train autonomous behaviors.

Responsible AI and trustworthiness

Encompasses all research geared toward ensuring AI systems are safe, and reliable. Core areas include LLM Safety and Security, Interpretability, Privacy Preservation, Unlearning (to remove specific data influence) and reliable decision-making through Causal Inference.

Model lifecycle and efficiency

Centers on the engineering challenges of operating models at scale. Involves improving the entire pipeline, including LLM Training, Inference and Optimization techniques like Model Inference Optimization to enhance the speed and resource efficiency of deployed systems.

Data and knowledge systems

Includes rigorous Data Curation, generation of reliable Synthetic Data, and the application of models to understand and predict real-world dynamics, such as User Behavior Modeling.

Our academia and science partnerships

We’re committed to partnering with leading research institutions to accelerate innovation, solve real-world problems, and deliver broad societal benefits. Learn more about our AI Academic Centers of Excellence, Partnerships and Community.

Work with us

• Explore our careers and join our world-class Applied Research team.
• Learn more about our Capital One PhD Fellows, our Applied Research Internship Program and our Data Science PhD Internship Program.