Teaching Modern AI to Understand the Subsurface: From Foundation Models to AI Agents

Description

What if AI could help uncover what lies beneath the Earth’s surface?

Seismic data is one of the most powerful tools for exploring the subsurface, but turning raw signals into real insight is complex, time-consuming, and often relies on expert intuition. This makes exploration geophysics the perfect place to ask a bold question: can modern AI transform how we do scientific data processing?

In this dynamic talk, you’ll see how cutting-edge visual and seismic foundation models are being used to tackle challenges like data interpolation and denoising. You’ll also get a glimpse into early experiments with large language models and tool-calling frameworks such as MCP, opening the door to more interactive, intelligent ways of working with data.

The session doesn’t stop there. It looks ahead to the future of AI in geoscience, from coding-agent-driven workflows to the potential of world models in exploration.

If you’re interested in AI, real-world research, or pushing the boundaries of what’s possible, this lecture will challenge your perspective and spark new ideas.

Lecturer Bio

Jiahua Zhao is a PhD candidate and Graduate Research Fellow at the Computation-based Science and Technology Research Centre of The Cyprus Institute, collaborating with TU Delft through the ENGAGE Marie Skłodowska-Curie programme.

His research focuses on high-performance GPU computing and applying foundation models to seismic data processing. He holds degrees in Oceanography and Geophysics, and has worked at leading institutions including Columbia University, Harvard, and KAUST.

His work brings together physics, high-performance computing, and AI to build faster, smarter workflows for exploration geophysics.

Notes

Free lunch will be provided for attendees.