- AI Systems - making frontier models fast and scalable.
- AI for Science and Biology - agents and foundation models across the scales of biology.
- Relational and Tabular Foundation Models - learning directly on the structured data that runs the world.
This workshop brings together Stanford researchers and industry partners to explore advances in data-centric AI, spanning tabular and relational foundation models, agentic systems, and applications in biology and science. Through research talks, case studies, and discussion, the workshop will highlight emerging ideas, foster new collaborations, and examine how frontier research creates value in real-world settings. Whether you're working in industry, academia, or startup land, there will be something to spark your curiosity and drive your work forward.
Registration
The Stanford Data-Centric AI Workshop (aka the Stanford Graph Learning Workshop) will be held on Tuesday, November 3, 2026, 08:30 - 18:00 Pacific Time.
The event will take place at Stanford University and will be live-streamed online on YouTube. Free registrations are available. Register now!
Schedule
Tentative agenda, subject to change.
| 08:30 - 09:00 | Registration & Breakfast | |
|---|---|---|
| 09:00 - 09:10 | Jure Leskovec (Stanford) | Welcome and Overview |
| 09:10 - 09:30 | Rishabh Ranjan (Stanford) | The Stanford Tabular and Relational (STAR) Ecosystem |
| 09:30 - 09:50 | Vignesh Kothapalli (Stanford) | Generating Large Scale Synthetic and Semi-Synthetic Datasets for Relational Foundation Models |
| 09:50 - 10:10 | Joseph Meyer (SAP) | RFMs & TFMs at SAP |
| 10:10 - 10:30 | Akihiro Nitta (NVIDIA) | Kumo Tabular: Pushing the Accuracy-Efficiency Frontier on Structured Data |
| 10:30 - 11:00 | Break | |
| 11:00 - 11:20 | Vignesh Kothapalli (Stanford) | Memorilla: Latent Semantic Memory for LLMs |
| 11:20 - 11:40 | Marcel Roed (Stanford) | Gigatoken: 1000x faster language model tokenization |
| 11:40 - 12:00 | Atindra Jha (Stanford) | M*: A Modular, Extensible, Serving System for Multimodal Models |
| 12:00 - 13:30 | Lunch & Poster session | |
| 13:30 - 13:50 | Yanay Rosen (Stanford) | Building Hierarchical Foundation Models for Biology |
| 13:50 - 14:10 | Kuan Pang (Stanford) | Learning What Data to Collect for Frontier Models in Biology |
| 14:10 - 14:30 | Fang Wu (Stanford) | Beyond Generation: Reasoning and Search for Protein Design |
| 14:30 - 15:00 | Break | |
| 15:00 - 15:20 | Moritz Schaefer (Stanford) | Agentic concept decomposition for scientific hypothesis generation |
| 15:20 - 15:40 | Tatsu Hashimoto (Stanford) | Data efficient language modeling |
| 15:40 - 16:00 | James Zou (Stanford) | Learning a physiology foundation model |
| 16:00 - 16:30 | Poster slam | |
| 16:30 - 18:00 | Happy Hour & Poster session | |
Posters
- Jiyuan Tan: Simons: Open Foundation Models for Zero-Shot Time Series Forecasting
- Huahai Yang: The Graph Is Not Given: A Declarative Semantic Layer for Relational Foundation Models
- Bingran You: FrontierPhysics: Evaluating Agents for End-to-End Physics Research
- Arjun Roy: TopoPrune: Robust Data Pruning via Unified Latent Space Topology
- Usama Bin Safeer: Dunbar based Structural encoding for Graph Transformers
- Venkat Dayala: Agentic Data Transformation and Mapping: Turning Heterogeneous Clinical Data into Foundation-Model-Ready Relational Databases
- Dijana Janjetovic: More Table per Token: Format Markup Does Not Pay Off in LLM Pretraining
- Rotem Ben-Shachar: AI-Powered Clinical Data Extraction for Rare Disease Diagnosis
- Luyang Zhang: Pick the Mixture, Not the Samples: Combinatorial Data Selection for LLMs
- Christos Ziakas: Fork-and-Flush: Escaping Idea Basins in Autoresearch Agents
- Harun Pirim: Constrained Neighbour Selection for Link Prediction on Noisy Knowledge Graphs
- Vassilis N. Ioannidis: Learning Enterprise Semantics for AI-Ready Knowledge Bases
- Tim Tsz-Kit Lau: Symmetry-Compatible Principle for Optimizer Design: Embeddings, LM Heads, SwiGLU MLPs, and MoE Routers
- Arpita Shanbhag: Training Size and Irrelevant-Feature Robustness in TabPFN
- Saptarshi Sengupta: On Link Prediction Using Node-Pair LLM Semantics and Structure in Temporal Text-Attributed Graphs
- Gajanan Chinchwadkar: Synapse: Multi-Model Graph Retrieval with Integrated Graph ML Execution
- Yiyuan Luo: Lend Me Your Eyes: Instruction-Aware Text Embeddings via Attention Relay
- Jinhee Lee: Surgical Vision-Language Model with AI-Generated Rationales
- Meghana Shanthappa: Combining Language-Derived and Walk-Text Features for Product Graph Classification
- Kristin Gleitsman: A Biologically Grounded Structural Causal Model Enables cfRNA-Specific In-Context Learning
- Kenny Peng: A Testable Theory of Atomic Features
- Mattias Akke: LLM sequential decision making under uncertainty in biochemical domains
- Aditya Shivakumar: FlowQuery: LLM-Based Text-to-SQL over Fragmented California Water Data for Farmers
- Michael Kim: Graph Neural Networks for Accelerated Materials Simulation
- Abhinav Sudhakar Dubey: Jailbreaking Open-Weight LLMs Via Random Embedding Perturbations
- Saksorn Ruangtanusak: Relational Deep Learning In Practice: From Enterprise Databases To Predictive Models
- Yuche Gao: CIDER-FM: Foundation Models for Causal Inference from Diverse Experimental Regimes
- Maksat Jazbay: SeisGeoSolver: Seismic modeling using graph neural networks and physics-attention mechanism
- Jongmin Sung: CombiCast: Human-in-the-Loop Multi-Agent Decision Support for Combination-Therapy Partner Selection
- Sirjan Ghimire: Noise is Not Always Noise: Task-Dependent Effects of Retrieval Noise on Abstractive QA and Knowledge Synthesis in Retrieval-Augmented Generation
- Jiajin Liu: GraphSearch: Agentic Search-Augmented Reasoning for Zero-Shot Graph Learning
- Haiming Tang: RelSports: A Multi-League Relational Benchmark for Sports Forecasting Across Leagues and Domains
- Zhenglun Kong: Structured Agent Distillation for Large Language Model Agents
- Utsav Gupta: When to Recheck: Discovery and Verification in a Simulated Protein Laboratory
- Shu Wan: Kov: A Causal Discovery Foundation Model via Amortized Stacking
- Sumanth Kumar Kotagudem: KORA: Verifiable Reasoning over Knowledge Graphs with Language Model Guidance
- Moritz Schaefer: Patient-level agentic analysis of deeply profiled cohorts uncovers CAR T cell therapy resistance
- Qiaoshi Jiang: Bridging Manufacturing and Clinical Outcomes with Interpretable Machine Learning: A Real-World CAR-T Cell Therapy Case Study
- Xi Xiong: Bayesian Offline Multi-Agent Reinforcement Learning with Latent World Models
- Wes Madrigal: KurveRSC: Task-Selected Relational Compression for Auditable Relational Prediction
- Sareh Nabi: When Does Structured Knowledge Help Neural Theorem Proving?
- Juwon Hong: Accurate Forecasts, Wrong What-Ifs: Foundation Models on Retail Data with Known Counterfactuals
- David Liu: Data Disaggregation for Group Personalization and Heterogeneity in Collaborative Filtering
- Agapi Rissaki: Database Views as Explanations for Relational Deep Learning
- Miguel Santiago Feles: Beyond Imitation: Knowledge Graph–Guided Language Model Distillation
- Yaguang Liu, Yulian Zhou, Shubhojeet Sarkar, Shengbo Guo: Exploring Forum with Generative Modeling
- Saad Naseem: AI-Guided High-Throughput Media Optimization for Enhanced Rare Earth Element Bioaccumulation
- Diganta Das: Multi-Objective Learning for Computational Antibody Sequence Co-Optimization: Generalization Under Sparse Binary Supervision
- Haoxin Liu: TIMERSI: Towards Self-Improved Agentic Deployment for Time-Series Foundation Models
- Dinesh Katupputhur Ramprasath: Support-Set Target Leakage in Relational Foundation Models during In-Context Learning: Impact, Detection, and Mitigation
- Karan Paresh: Ownership Is the Missing Declaration: Deciding Relational Target Leakage from the Schema
- Sukwon Yun: Graph-of-Agents: A Graph-based Framework for Multi-Agent LLM Collaboration
- Aarav Sekhri: Toward Parameter-Efficient and Trustworthy Transformers for Stress Detection in Reddit Posts
- Anu Vellore: matplotgym: Can Language Models Turn Data into Publication-Quality Figures?
- Kargi Chauhan: Did the AI Discover It, or Remember It? A Placebo Test for AI Co-Scientists
- Arefeh Yavary: GENESIS: Autonomous Drug Discovery Where World Models Meet Adaptive Agents and Personalized Patient Digital Twins
Organizing committee
Lata Nair, Charilaos Kanatsoulis, Moritz Schaefer, Michael Bereket, Kuan Pang, Jure Leskovec