
Matt
@mmoderwell
Building @ourofoundation. Working on materials discovery (rare-earth-free permanent magnets, superconductors) and the creator economy. AMDG.
你可能會喜歡
A materials scientist agent is in the works! Going to use DSPy to optimize my way to a rare-earth-free permanent magnet discovery. The past few months I'd been adding packaged materials science APIs and workflows to Ouro. These are fun to use manually but their real power shows…

Getting back to the main reason I started Ouro - give technical creators a platform for their work. Especially creators that want to earn from their work. And while it doesn't matter what you create, you still may ask: Was it interesting? Was it useful? Now there's a place to…

One of my goals for Ouro has been to help creators and researchers get their work out in front of more people. SEO has been a major part of this plan from the start. Even the the way the pages are rendered and the metadata we add to the HTML head all play a part. I just added a…
Last night over Lake Michigan. Looked like a sunrise it was incredible.

🪢 Innovation out of necessity. This is why rare-earth-free permanent magnets have been my top priority. My team and I are using AI to accelerate the discovery of hard magnets that have no dependencies on sensitive supply chains. 🇨🇳 Make it hard to get rare-earths. 🇺🇸 Find a…
CHINA JUST WEAPONIZED THE ENTIRE RARE EARTH SUPPLY CHAIN get used to reading chinese MOFCOM bulletins because they're writing american industrial policy now. announced today: any product containing >0.1% chinese rare earth materials needs beijing's approval before re-export to…

Still get excited every time I see Ouro show up in SERP or a ChatGPT response. I put a lot of work into SEO early on so it's nice to see it's finally paying off. If you have a scientific model or dataset to share, Ouro helps you get it out there.

Things turn a turn when I turned on gpt-5-high. It finally told me that what I was trying to do wouldn't work. It wasn't a code challenge. No, it was that the model I was working with did not have the physics I needed. I didn't believe it at first. I had 80 hours of work with…
The past week straight I've been vibing a magnetocrystalline anisotropy energy predictor that could run in @modal so users can use it whenever they need it. $100 in Cursor credits and almost 200 million tokens later, let me reflect on what I've learned.


Turns out the standard orbital sets are not all the same across DFT softwares. I’ve been trying to run some lightweight DFT in Modal for a new API and part of this process involves taking DFT outputs as inputs to a pretrained model. That model was trained OpenMX data. I thought…
Can’t wait to see what you guys will discover. Happy building 🫡
Today, @ekindogus and I are excited to introduce @periodiclabs. Our goal is to create an AI scientist. Science works by conjecturing how the world might be, running experiments, and learning from the results. Intelligence is necessary, but not sufficient. New knowledge is…

After some conversations and reflections on my choice of mutations for the AI materials discovery scientist, I'm starting to see how to improve. While the results were very promising, I was disappointed by the P1 symmetry that often came out. Initial set of mutations: -…
United States 趨勢
- 1. No Kings 788K posts
- 2. Vandy 6,368 posts
- 3. Diego Pavia 1,751 posts
- 4. Dork Cult Protest Day 34.6K posts
- 5. #AnteX N/A
- 6. Brian Kelly 1,244 posts
- 7. Vanderbilt 5,230 posts
- 8. Arsenal 126K posts
- 9. Nuss 2,733 posts
- 10. #GoBlue 1,929 posts
- 11. Cole Sullivan N/A
- 12. Shula 1,345 posts
- 13. Duke 72K posts
- 14. Semaj Morgan N/A
- 15. Beamer 1,563 posts
- 16. Fulham 38.8K posts
- 17. Saka 36.5K posts
- 18. Zavion Thomas N/A
- 19. Sherrone Moore N/A
- 20. #Sooners 1,055 posts
Something went wrong.
Something went wrong.