A geosciences blog
Earlier this year, I was invited to give a talk at the Bay Area Geophysical Society. I ended up presenting a set of ideas I ve been thinking about regarding the current state of AI in geophysics, and where I think it s going to lead in the future. (I also presented this work at an AI [ ]
This post is part of GP³, an ongoing series on Gaussian Processes and their applications in Geology and Geophysics. The full notebook and source code are available for you to explore on GitHub! A few months ago, the PyTorch blog announced that NeuralOperator has now joined the PyTorch ecosystem! Neural operators are one instance of [ ]
2025 was a blockbuster year for both AI and critical minerals! Pretty much every other week either we got news of a new discovery or governmental initiative, or an AI company releasing a new model or product. To break through the noise and try to understand the trends of the year, I thought it would [ ]
Earlier last month I went to a Google Gemini vibe coding hackathon in SF, hosted by Cerebral Valley. Vibe coding, for the unfamiliar, is the practice of heavily using AI-assistant tools for software development, letting you focus on the high-level structure and results rather than the exact details of implementation. The theme of the hackathon [ ]
At Terra AI, I have a great collaborator who describes himself as optimization-brained, and always comes up with the best strategies for optimization problems, whether he s coming up the with best linear algebra way of performing a calculation, or he s thinking about the best way to apply a set of constraints. This is in contrast [ ]
Last weekend I had the pleasure of participating in the official OpenAI GPT-5 hackathon in San Francisco! The event was thrilling, it was amazing to see so many passionate people trying out new things with GPT-5. I figured I d write up what I developed in my weekend project, as well as gave some thoughts on [ ]
Gaussian processes (GPs) are powerful tools for modeling and quantifying uncertainty over functions that are smooth and continuous. When used for regression, they are one of the only methods that allows you to do Bayesian inference in infinite-dimensional spaces, without either: resorting to approximate inference methods like RTO-TKO [1], or truncating the problem ab initio [ ]
Recently in my work I ve been reading papers in areas of computer science (mostly machine learning), and I ve noticed how the titles are markedly different to those I m familiar with in Earth sciences. Titles are attention-grabbing, feeling playful and almost whimsical. However, there seems to be a great deal of repetition, becoming cliché at times. [ ]
This post is a summary of a series of experiments I recently conducted, see the following gist: link. In this post, I explore the flexibility of Gaussian process (GP) kernels in GPyTorch. Specifically, I m interested in their ability to learn appropriate length-scales inherent in spatial data. I think Gaussian processes are great! They are a [ ]
After taking an unintended break from blogging, I m back! Quick update, I decided to leave my fellowship position at IGPP, and now I m a senior computational scientist at Terra AI! I plan to write a longer blog post about this when I have the time, and talk about career choices, and the things I learned [ ]