Announcing Capital One Fellows for academic year 2026-2027

We’re thrilled to introduce the Capital One Fellows award recipients for the 2026-2027 academic year. The Capital One Fellows program is an annual fellowship for doctoral students, designed to equip emerging researchers with the skills and knowledge needed to excel in the field of AI, machine learning and data science. This exceptional group of researchers is working alongside leading faculty to tackle complex challenges and push the boundaries of their field. (Learn more about our academic partnerships.)

Here’s a closer look at this year’s Capital One Fellows and their upcoming projects.

Columbia University

Tao Long
Project Title: Vibe-Working: Agentic Infrastructure for Humans to Become AI Supervisors for General Knowledge Work
Faculty Advisor: Dr. Lydia Chilton

Tao Long is a computer science PhD student at Columbia University, advised by Professor Lydia Chilton. Addressing human-AI interaction, Tao’s research explores how humans collaborate with generative AI systems and AI agents over time, focusing on making AI tools more usable, useful, trustworthy, reliable and seamlessly integrated into everyday productivity practices. Specifically, Tao builds human-AI and agentic systems that reduce cognitive and temporal effort for challenging or complex tasks; offload work to AI while maintaining human ownership and authenticity; and fit naturally into the existing processes of writers, developers, designers, event organizers and many other communities. Before starting their PhD, Tao earned a BS summa cum laude from Cornell University.

Jinjun Peng
Project Title: Secure CodeLLM with Reasoning
Faculty Advisor: Dr. Baishakhi Ray

Jinjun Peng is a PhD student in computer science at Columbia University, advised by Professors Baishakhi Ray and Junfeng Yang. His research has centered on code language models, including SWE-Spot, which builds small repository-expert models through repository-centric learning; CWEval, an outcome-driven benchmark evaluating both the functionality and the security of LLM-generated code; and SemCoder, which trains code language models with comprehensive semantic reasoning. His current research interest is continual learning: how AI agents can effectively and efficiently acquire new knowledge and adapt to evolving environments.

Ved Sriraman
Project Title: Understanding and Stabilizing Reinforcement Learning Dynamics in Post-Training Language Models
Faculty Advisor: Dr. Adam Block

Ved Sriraman is a computer science PhD student at Columbia University, advised by Professor Adam Block. His research focuses on improving the capabilities and reliability of LLMs through post-training and test-time algorithms. Combining theoretical analysis with empirical investigation, Ved develops algorithmic interventions that come with provable guarantees while remaining relevant to modern language-model training. More broadly, he is interested in uncovering fundamental mathematical principles of learning in real-world environments and distilling them into methods that enable language models to reason. Before starting his PhD, Ved earned a BS in computer science from Cornell University.

Hideaki Takahashi
Project Title: Proactive Defenses for Large Language Models

Hideaki Takahashi is a PhD student in computer science at Columbia University, researching software security, cryptography and AI agents. He is also an active ethical hacker and has discovered over 100 zero-day vulnerabilities in widely used Web3 and AI-agent projects.

University of Illinois Urbana-Champaign

Yizhuo Chen
Project Title: Utility-Preserving Data Sanitization for Generative AI Model Development
Faculty Advisor: Dr. Tarek Abdelzaher

Yizhuo Chen is a PhD candidate in computer science at the University of Illinois Urbana-Champaign, advised by Professor Tarek Abdelzaher. His research centers on trustworthy machine learning and foundation models for sensing, with a particular focus on privacy-preserving data analytics and the safe use of sensitive data in generative AI. Yizhuo’s work has appeared at venues such as ICML and CVPR, and he has collaborated with industry research labs at JPMorgan Chase and Amazon. He received his BSc from Zhejiang University in 2021.

Hyeonjeong (Amber) Ha
Project Title: Toward Creative and Reliable Large Language Models
Faculty Advisor: Dr. Heng Ji

Hyeonjeong “Amber” Ha is a third-year PhD student at the University of Illinois Urbana-Champaign, advised by Professor Heng Ji. Her research enhances the visual perception of multimodal LLMs through fine-grained, structured understanding, aiming to bridge humanlike perception and trustworthy reasoning for more reliable real-world applications. She has published as first author in top-tier conferences including ACL and NeurIPS and actively contributes to the research community as a reviewer for leading venues.

Jeonghwan Kim
Project Title: Enhancing Real-World Grounding through Textual-Visual Representation Interleaving in Multimodal Large Language Models
Faculty Advisor: Dr. Heng Ji

Jeonghwan Kim is a PhD candidate in computer science at the University of Illinois Urbana-Champaign, advised by Professor Heng Ji. His research focuses on multimodal foundation models that bridge fine-grained visual perception, reasoning and embodied intelligence. Prior to his work on multimodal AI, Jeonghwan conducted research in natural language processing, including multi-hop question answering, retrieval-augmented generation and numerical reasoning. His work has appeared in leading venues such as NeurIPS, ICLR, ACL, EMNLP, NAACL and CVPR, including a NeurIPS 2025 Spotlight paper on part-level visual understanding in large multimodal models.

University of Virginia

Zhenyu Lei
Project Title: Towards Efficient, Reliable, and Multimodal AI Reasoning in Financial Services
Faculty Advisor: Dr. Jundong Li

Zhenyu Lei is a PhD student in electrical and computer engineering at the University of Virginia, advised by Professor Jundong Li. Before attending UVA, he earned his BS in physics (honors graduate) from Xi’an Jiaotong University. His research centers on making LLM reasoning more efficient and reliable. Specifically, he works on reasoning distillation, which compresses powerful models into smaller ones that retain strong reasoning ability, and on reasoning editing, which corrects failures in model behavior and enables LLMs to reason reliably over their lifetime. Building on this foundation, Zhenyu is excited to extend his work to the financial domain, where efficient and trustworthy reasoning is critical for real-world decision-making. He has published over 20 papers at leading venues including ICLR, AAAI and ACL, with three oral presentations.

Zhepei Wei
Project Title: Toward Trustworthy and Efficient Agentic AI
Faculty Advisor: Dr. Yu Meng

Zhepei Wei is a PhD candidate in the computer science department at the University of Virginia, advised by Professor Yu Meng. He has held research positions at Microsoft Research, Meta and Amazon, working on large language models. His first-authored research papers have been published in top-tier venues in the fields of machine learning and artificial intelligence (e.g., NeurIPS, ICML, ICLR), with over 1,400 citations on Google Scholar. Zhepei is also a recipient of the UVA Copenhaver Charitable Trust Bicentennial Fellowship and the John A. Stankovic Outstanding Graduate Research Award.

Hanzhang (Mia) Yuan
Project Title: Integrative Decoding for Reliable Large Language Model Reasoning in Financial Services
Faculty Advisor: Dr. Sheng Li

Hanzhang “Mia” Yuan is a fourth-year PhD student in data science at the University of Virginia, advised by Dr. Sheng Li. Her research focuses on improving the reasoning and reliability of large language and multimodal models. As a Capital One Fellow, she is extending this work to financial tabular reasoning, with an emphasis on helping models interpret structured information and perform complex numerical reasoning more faithfully. Mia’s broader goal is to develop trustworthy AI systems that can support reliable decision-making in real-world financial applications.

Ding Zhang
Project Title: Towards Multimodal Graph-Language Reasoning
Faculty Advisor: Dr. Chirag Agarwal

Ding Zhang is a second-year PhD data science student at the University of Virginia, advised by Professor Chirag Agarwal. His research focuses on multimodal graph-language reasoning — modeling graph-structured data paired with textual and other modalities using graph representation learning techniques, such as GNN models. Additionally, as AI systems become increasingly integrated into critical decision-making processes in areas such as healthcare, finance and public policy, it is essential that these systems are not only accurate but also interpretable and reliable. Building models that can clearly explain their predictions is crucial for fostering user trust and ensuring ethical deployment. This area is the key to bridging the gap between technical innovation and real-world, responsible adoption of AI.

University of Southern California

Amin Banayeeanzade
Project Title: Unlocking Diverse Thinking and Planning in AI Agents
Faculty Advisor: Dr. Sai Praneeth Karimireddy

Amin Banayeeanzade is a PhD student in computer science at the University of Southern California. He studies the trustworthiness of AI systems with a particular focus on diversity in large language models and agentic systems. Amin completed his master’s and undergraduate degrees at Sharif University of Technology.

Xinyue Cui
Project Title: Multi-Signal Reward Modeling for Robust and Trustworthy AI
Faculty Advisor: Dr. Swabha Swayamdipta

Xinyue Cui is a third-year PhD student at the University of Southern California, advised by Professor Swabha Swayamdipta. Her research focuses on advancing AI safety and trustworthiness through a data-centric lens, investigating how the quality, integrity and diversity of training data shape model reliability, robustness and susceptibility to manipulation. Prior to her PhD, Xinyue received her MS in computer science from USC and her BS in computational mathematics from the University of California-Los Angeles.

Zarif Ikram
Project Title: Belief-Weaver: Probabilistic Belief Transformers for Synergizing Language, Knowledge, and Rich Feedback
Faculty Advisor: Dr. Paria Rashidinejad

Zarif Ikram is a second-year PhD student in computer engineering at the University of Southern California, advised by Professor Paria Rashidinejad. His research interests lie at the intersection of world models and scalable oversight, with a major theme being the principled understanding of “why” in training foundation models. Zarif earned a BS from Bangladesh University of Engineering and Technology in 2025 and interned at Microsoft Research in 2026. 

Md Abrar Jahin
Project Title: FinSemKG: A Semantic Mediation Layer for Grounded, Auditable LLM Question Answering over Financial Tables
Faculty Advisors: Dr. Craig Knoblock and Dr. Jay Pujara

Md Abrar Jahin is a second-year PhD student in computer science at the University of Southern California, advised by Professors Craig Knoblock and Jay Pujara. He is also an AI researcher with the Center on Knowledge Graphs at USC’s Information Sciences Institute. Abrar’s research focuses on structured and trustworthy AI, particularly knowledge representation, semantic table understanding, geometric deep learning and grounded LLM systems. As a Capital One Fellow, he is developing FinSemKG, a semantic mediation layer for grounded, provenance-aware and auditable LLM question-answering over heterogeneous financial tables. He also contributes to projects funded by DARPA/USGS and AFRL involving scientific knowledge extraction, knowledge graph construction and critical mineral discovery.

Abrar has authored more than 20 peer-reviewed publications that have received over 700 citations. His distinctions include the Viterbi Graduate School Fellowship and two Highly Commended distinctions from the Global Undergraduate Awards, placing his undergraduate research in the top 10% globally. He also serves as a program committee member and reviewer for leading AI and machine learning venues, including AAAI, ICDM, NeurIPS, ICML and TMLR, and received an ICML 2026 Gold Reviewer Award. Before joining USC, he earned his BSc in integrated process excellence from Khulna University of Engineering and Technology and served as a visiting researcher at the Okinawa Institute of Science and Technology in Japan.

Asal Mehradfar
Project Title: Uncertainty Quantification for Multi-Agent LLM Systems: Decomposition and Failure Localization
Faculty Advisor: Dr. Salman Avestimehr

Asal Mehradfar is a fourth-year PhD candidate in electrical engineering at the University of Southern California, advised by Professor Salman Avestimehr. Her research focuses on developing AI and machine learning methods for engineering systems and scientific discovery. She is particularly interested in building reliable, domain-informed learning and generative frameworks that integrate data, scientific knowledge and human expertise. Her work spans financial market modeling, electronic design automation and biomedical systems.

Kien Nguyen
Project Title: Dynamic Schema-guided Monte Carlo for Text-to-SQL Agents on Massive Databases
Faculty Advisor: Dr. Paul Bogdan

Kien Nguyen is a fourth-year PhD student in computer engineering at the University of Southern California, advised by Professor Paul Bogdan. His prior research focused on machine learning and graph representation learning, and he is currently interested in leveraging graph learning techniques for LLM agent research, including self-evolving and multi-agent systems. In collaboration with Capital One, Kien is developing advanced agent systems for data analytics and decision-making tasks across businesses. He has published his work in and served as a reviewer for prestigious venues such as AAAI, WWW and KDD.

Shahab Sepehri
Project Title: XFinAgent: Financial Analyst Agents with Sparse-Probe Control, Multi-Agent Explainability, and Robust Evaluation
Faculty Advisors: Dr. Mahdi Soltanolkotabi and Dr. Stephen Tu

Shahab Sepehri is a fourth-year PhD candidate advised by Professors Mahdi Soltanolkotabi and Stephen Tu at the University of Southern California. His research focuses on developing new algorithms for multimodal reasoning in vision-language and diffusion models, enabling them to better integrate, generate and reason across modalities. Shahab also works on time-series foundation models, developing more effective and efficient tokenization schemes for diverse temporal data.

Yuxin Yang
Project Title: Query-Adaptive Graph Traversal and Persistent Investigation Memory for Financial Knowledge Extraction
Faculty Advisor: Dr. Viktor Prasanna

Yuxin Yang is a fourth-year PhD student in computer science at the University of Southern California, advised by Professor Viktor Prasanna. Her research focuses on building reliable and efficient LLM agents, with particular emphasis on controlled test-time scaling—enabling agents to adaptively decide how to search, reason, verify and manage context during long-horizon tasks. In collaboration with Capital One, Yuxin developed SPARC-RAG, an agentic retrieval-augmented generation framework that dynamically allocates computation across sequential and parallel search while consolidating evidence in a unified context state. Her broader research spans post-training, deep research and coding agents, as well as graph learning, with the goal of making AI systems more capable, efficient and trustworthy in complex decision-making settings. Before joining USC, she received her Bachelor of Engineering degree from Tsinghua University.