AI RESEARCH · PHD CANDIDATE
Learning from language.
Reasoning with structure.
I research foundation models, LLMs, and graph learning for relational data and temporal sequences—connecting research with real-world problems in finance, enterprise AI, and healthcare.
01 / INDUSTRY EXPERIENCE
AI research at SAP Labs and Capital One, spanning relational foundation models and large-scale fraud detection.
Palo Alto, CA
AI Research PhD Associate Intern
Researching graph, relational, and tabular foundation models for multi-table data. Developing parameter-efficient relational adaptation, relational in-context learning, and graph-based retrieval to generalize across datasets and tasks.
AUG 2026 – PRESENT
New York, NY
Principal Data Science Intern · AI Foundations
Developed HERMES, a dual-scale relational Graph Transformer combining heterogeneous temporal neighborhoods with ecosystem-level context for consumer identity and fraud detection.
JUN – AUG 2026
San Jose, CA
Principal Data Science Intern · AI Foundations
Built ATLAS, reframing account takeover detection as directed temporal graph learning. Combined causal message passing, inductive neighbor sampling, and lag-aware supervision on a graph with 100M+ nodes and 1B edges.
JUN – AUG 2025
02 / FEATURED RESEARCH
My work connects the semantic capabilities of language models with the relational structure of graphs. Selected work on language models, graph learning, and knowledge augmentation.
Import What You Need: Learning When and How to Augment EHR Graphs with External Knowledge
Learning when to import external knowledge—and whether to enrich features, modify graph structure, or skip augmentation—using a budget-aware reinforcement learning policy.
RePrompT: Recurrent Prompt Tuning for Integrating Structured EHR Encoders with Large Language Models
Recurrent soft prompts connect structured EHR encoders to LLMs, preserving longitudinal patient context and incorporating population-level patterns.
03 / PUBLICATIONS
Manuscript under review
Manuscript under review
Manuscript under review
EMNLP 2026 · Main Conference Accepted
04 / TALKS & POSTERS
Presentation slides and posters exploring the methods, experiments, and ideas behind my research.
EMNLP 2026
REFINE: LLM Refinement over Budgeted Text-Attributed Graphs for Personalized Medical Concept Representation
Research poster · PDF ↗
ACL 2026
Text-Attributed Knowledge Graph Enrichment with Large Language Models for Medical Concept Representation
Talk slides · PDF ↗
CIKM 2025
Multi-Ontology Integration with Dual-Axis Propagation for Medical Concept Representation
Talk slides · PDF ↗
NeurIPS 2025
Spatio-Temporal Directed Graph Learning for Account Takeover Fraud Detection
Research poster · PDF ↗
CIKM 2023
Contrastive Learning of Temporal Distinctiveness for Survival Analysis in Electronic Health Records
Talk slides · PDF ↗
Survival model robustness
SurvAttack: Black-Box Attack on Survival Models through Ontology-Informed EHR Perturbation
Research poster · PDF ↗05 / CONFERENCES & MILESTONES
Presentations, conversations, and milestones from my research journey.
06 / BACKGROUND
I’m Mohsen Nayebi Kerdabadi, a final-year PhD candidate in Computer Science at the University of Kansas, with 5+ years of AI research experience.
My research spans knowledge-grounded LLMs, graph and relational foundation models, adaptive retrieval, and temporal representation learning. I develop methods that bring language, structure, and context together—and evaluate them on problems where those connections matter.
I have been a Research Assistant at KU since January 2023. My industry research includes SAP Labs and Capital One’s AI Foundations team.
University of Kansas
University of Kansas
Isfahan University of Technology
Reviewer for NeurIPS, ICLR, KDD, CIKM, SDM, IJCAI, TKDD, and the Journal of Biomedical Informatics.
Teaching at KU includes Data Mining and Advanced Data Science.
LET’S CONNECT
Foundation models. Structured data. Real-world AI.
