Moderator: Neeraj Kumar, Pacific Northwest National Laboratory
Vivek Natarajan, Google DeepMind
AI for Science: Foundations and Frontiers is a hands-on tutorial designed to equip researchers with practical skills and conceptual grounding in the application of large-scale AI models to scientific challenges.
The program covers key components of the AI model lifecycle: from distributed strategies for pre-training generative models to fine-tuning techniques for domain-specific tasks using models like LLAMA-70B and Stable Diffusion.
Participants will also learn to analyze and optimize performance through workload profiling with PARAVER, and to build intelligent scientific workflows using Retrieval-Augmented Generation (RAG) and agent-based approaches. The tutorial concludes with real-world case studies across disciplines—biology, climate, physics, chemistry—highlighting lessons learned from deployment and emerging trends such as simulation models and neural-symbolic systems.
Participants will develop a practical understanding of large-scale AI model development, including:
With exposure to real-world scientific applications and current research frontiers in AI for science.
Instructors:
Experts from Argonne, ORNL, PNNL, CINECA, BSC, and others.
Evaluation of AI Model Scientific Reasoning Skills is a hands-on tutorial designed to equip researchers with practical skills and conceptual grounding in the application of LLMs to scientific challenges.
Large Language Models (LLMs) are becoming capable of solving complex problems while presenting the opportunity to leverage them for scientific applications. However, even the most sophisticated models can struggle with simple reasoning tasks and make mistakes.
This tutorial focuses on best practices for evaluating LLMs for science applications. It guides participants through methods and techniques for testing LLMs at basic and intermediate levels. It starts with the fundamentals of LLM design, development, application, and evaluation while focusing on scientific application. Participants will also learn various complementary methods to rigorously evaluate LLM responses in benchmarks and end-to-end scenario settings. The tutorial features a hands-on session where participants use LLMs to solve provided problems.
Participants will learn the principles and approaches for the use of LLMs as scientific assistants and how these can be evaluated with respect to scientific knowledge and reasoning skills, such as:
Instructors:
Franck Cappello, Sandeep Madireddy, Neil Getty (Argonne), Javier Aula-Blasco (BSC)
Session 1: Plenary session with all Tutorial and Hackathon participants: Foundations in AI for Science
Session 2: Case Studies and Emerging Frontiers in AI for Science
Session 2: Use Cases and Basic Evaluation Techniques
Session 3: Fine-tuning Techniques: From Theory to Practice
Session 3: Advanced Evaluation Techniques
Session 4: Building RAG-based Workflows
Session 4: Hands On
This is a hands-on tutorial designed to equip researchers, students, engineers, and scientific leaders with practical skills for the use of AI models and systems to accelerate planning, inquiry, coding, and other common tasks.
This tutorial will demonstrate how computational scientists can effectively harness AI-powered tools across the research lifecycle to increase their productivity. Attendees will learn to generate novel research ideas and hypotheses using agents for Deep Research and Idea Generation. We then cover structuring comprehensive research plans with AI assistance.
For implementation, attendees will see how to efficiently port, develop, and optimize code using tools like Google’s Gemini Code Assist and CLI, alongside advanced optimizers such as AlphaEvolve. While this tutorial will use Google technologies for the examples (and attendees will be given accounts to access them), the core principles and strategies are designed to be portable, enabling scientists to effectively use any comparable AI tool in their own scientific endeavors.
Session 5: Foundation Models for Computational Fluid Dynamics and For Scientific Data Compression
Session 1: LLM Refresher + Deep Research & Idea Generation
Graph Foundation Models for Materials Discovery + Unified Framework for Scalable and Efficient Vision Transformer Models: A Case Study with ORBIT
Session 2: Coding Faster & Better (Usually) + AI-enabled Science Applications
TPC26 is produced by TCI Media, home of the Wire Publications:
© 2026 TCI Media.
TPC25 is produced by TCI Media, home of the Wire Publications:
© 2025 TCI Media.
