Hi! I'm Lyna :)
I'm building ways to distribute wicked fast, terrifyingly accurate AI-based weather forecasts at WindBorne Systems. In past lives, I've been an acquired founder, an AI researcher, and climate policy fellow at NOAA.
I'm looking for my next obsession in the realm of building useful AI for humans. Grab a time or contact me if you have cool ideas!
note from 2025.03.10 — I'm heading to Singapore in April for ICLR 2025. Reach out if you'll be there, coffee's on me!
current thoughts
a casual dump of thoughts/questions in my head, usally no more than a few months stale
-
what learnings can we discover & operationalize about ai weather model trustworthiness from the scenarios under which forecasts are unphysical?
-
what do llms teach us about the human cognitive experience? are llms using the same neural representations of the world? what do the ovverlaps/differences tell us about what it means to "learn" or "think" or "feel"?
-
how can i sustainably create tangible artistic artifacts of my emotional experiences?
previously
not a resume, instead a story in bullet points of interesting past chapters
coming soon!
archive of previous thoughts
-
2024.03 — do ai x weather models learn physics? how is this correlated with model performance, particularly in relation to physics-based numerical weather prediction systems?
things i do
different ways i find meaning
- research: exploring the intersection of AI, humans, and complex systems
- social dance: learning a new language for expression and connection
- other labors of love: building miscellaneous projects that obsess me
research
at the highest level, i'm excited to better understand or benefit humans through their interactions with complex systems—like the environment, or markets, or LLMs.
more specifically, i'm excited about ai for science, particularly foundation models for the earth. that said, my broader research interests have led me to explore a pretty odd set of corners, like: the time dependency of carbon credits, physics-based interpretability of ai for numerical weather prediction, and what we can learn of the inner workings of human brains from LLMs.
conferences * denotes equal contribution
-
WeatherMesh-3: Fast and Accurate Operational Global Weather ForecastingICLR 2025 Spotlight Presentation in Tackling Climate Change with AI
-
Extreme Weather BenchAMS 2025 Oral Presentation [proc]
-
Towards Physically Interpretable and Consistent AI-Driven Numerical Weather Forecasts
-
CloudTracks: A Dataset for Localizing Ship Tracks in Satellite Images of CloudsAGU 2024 Oral Presentation [proc]
-
Enhancing Numerical Weather Prediction with Artificial Intelligence: Activities of the NOAA AI for Numerical Weather Prediction (AI4NWP) TeamAGU 2024 Oral Presentation [proc]
-
Navigating Data-Centric AI Challenges in Earth and Space Sciences: Focus on Climate and Remote SensingAGU 2024 Invited Presentation [proc]
not conferences
-
Nature in the balance: What companies can do to restore natural capital
social dance
coming soon!