Research scientist & computational chemist @BerkeleyLab using HT DFT workflows, machine learning, and reaction networks to model complex reactivity.
Joined September 2020
- Tweets255
- Following659
- Followers1.3K
- Likes1.1K
Pinned Tweet
The Open Molecules 2025 dataset is out! With >100M gold-standard ωB97M-V/def2-TZVPD calcs of biomolecules, electrolytes, metal complexes, and small molecules, OMol is by far the largest, most diverse, and highest quality molecular DFT dataset for training MLIPs ever made 1/N
Sam Blau retweeted
1. Just a head's up to those not paying close attention: it's a semiconductor, not superconductor.
2. I hope these "announcements" don't become the norm. If I posted DFT calcs of all the things I thought I "found" before digging into it, you'd laugh. This isn't much better, IMO.
For decades, researchers have sought materials that sort electrons by spin while their magnetism cancels.
In 3 days, 90+ Opus 5.5 agents helped us uncover two room-temperature magnetic semiconductor candidates in simulations: YBaMnFeO₅ and KV[Cr(CN)₆].
KV[Cr(CN)₆] was synthesized back in 1999. Its predicted ability to sort electrons by spin appears to have been hiding in plain sight for 27 years.
Sam Blau retweeted
Introducing Zatom-1, the first end-to-end, fully open-source foundation model for 3D chemistry! This was a great collaborative effort with many brilliant scientists. I'm grateful to have played a small part.
Paper: arxiv.org/abs/2602.22251
Code: github.com/Zatom-AI/zatom
ALT Architecture of the Zatom-1 foundation model for 3D molecules and materials. A unified Transformer trunk processes (noisy) atomic, structural, and conditional inputs. The model features two training stages: (1) multimodal flow pretraining using the final trunk layer L’s representations to create new molecules or materials; and (2) multi-task finetuning of additional downstream task backbones and heads that use a specified trunk layer K’s representations to predict properties, energies, and forces.
Sam Blau retweeted
Open Molecules 2025 just got even better! We’re pleased to announce today that we significantly expanded OMol25 with 40M new datapoints. We also present our collaboration with Lawrence Livermore National Lab and Lawrence Berkeley National Lab on the Open Polymer 2026 dataset.
Out in @NatComputSci: A roadmap for inverse design of nanomaterial heterostructures via HT data gen -> representation dev -> heteroGNN training -> gradient-based global opt! w/ @EmoryChanNano
nature.com/articles/s43588-0…
🔓rdcu.be/eTH72
@BerkeleyLab
@molecularfndry
@MITChemE
Emory and I also wrote a higher level summary of the work: nature.com/articles/s43588-0…
🔓rdcu.be/eUVIw
Sam Blau retweeted
An accompanying Research Briefing is now available for this paper! nature.com/articles/s43588-0…
🔓rdcu.be/eUVIw
Sam Blau retweeted
📢@EmoryChanNano, @SamMBlau and colleagues develop a DL approach for optimizing the nonlinear optical properties of core-shell upconverting nanoparticles, uncovering photophysical design rules and a roadmap for DL in nanoscience. nature.com/articles/s43588-0…
🔓rdcu.be/eTH72
Sam Blau retweeted
At this point the AI for Science community should stop focusing on achieving "state-of-the-art” on datasets like QM9 & MD17: chasing small improvements on these outdated datasets is scientifically meaningless. It's like telling vision researchers to ditch internet-scale and go back to benchmarking on MNIST/CIFAR10
Sam Blau retweeted
If you are a PhD student interested in AI for chemistry/biology/materials science, come join us! We work on a range of problems from generative modeling for inverse design and sampling to interatomic potentials and electronic structure prediction. Application link below.
Sam Blau retweeted
We are actively hiring! We are a small team in Boston building real-time AI systems to transform how advanced materials are engineered.
If you live at the intersection of science (especially materials), AI/ML, and/or software engineering, please consider applying.
🚀 We’re hiring!
Atomscale is looking for a talented Applied AI/ML Engineer to join our team in Boston, MA.
If you’re excited by the idea of applying machine learning to real-time materials processing and enabling atomic-scale precision in manufacturing, we’d love to talk. This role bridges AI, high-performance computing, and advanced materials science—helping shape the future of semiconductor and nanotech innovation.
Join us at the intersection of AI and materials, where your models directly impact how next-generation materials are discovered and produced.
👉 Learn more and apply here: atomscale.ai/careers/applied…
#hiring #machinelearning #artificialintelligence #materials #deeptech #startup #atomscale #Boston
I'm hiring postdocs @BerkeleyLab to drive cutting-edge research involving MLIPs, high-throughput workflows, chemical reaction networks, generative models, and open-source software dev. Full position description + application here: forms.gle/zePBZDmciXezyfe39 #Chempostdoc #AI4Science
Sam Blau retweeted
Quick update on the OMol25 Electronic Structures dataset we released two weeks ago with our partners at @metaai. Access to the dataset is now dramatically simpler! Simply go to this page, join a Globus group, and access the data.
🐻Please bear with us as this is a uniquely large dataset (~500 TB of open data). We've seen a lot of interest in this dataset, and are still working to optimize performance of the endpoint.
Link: materialsdatafacility.org/sp…
Phenomenal work by Aditi and her team! So cool to see what OMol25 is enabling - and it’s still really just the start!
1/ Can molecular AI move past hard-coded Graph Neural Networks and embrace scalable Transformers that discover molecular structure on their own?
We demonstrate that you can train a 1B parameter Transformer model without any graph priors or physical inductive biases.
And surprisingly, not only can it maintain competitive performance under equal compute on the Open Molecules 2025 dataset… it’s faster than a 6M parameter equivariant GNN, and exhibits scaling laws that don’t saturate. We use this as a starting point to investigate emergent internal representations, and find that it adaptively discovers molecular structure!
Check out the interactive demo on our website: tkreiman.github.io/projects/…
And our paper: arxiv.org/abs/2510.02259
In collaboration with @tobykreiman, @YutongBAI1002, Fadi, Elizabeth, and @EricQuCal.
Here’s a video showing how the Transformer learns distance-aware attention patterns (purple gradient) that adapt to atomic environments 👇
Sam Blau retweeted
🔥 Today we announce the Meta OMol25 Electronic Structures Dataset - 500 TB of molecular data in collaboration with @mshuaibii and team at @AIatMeta. We envision a future where researchers can rapidly design molecules and peptides to treat diseases, discover catalysts to revolutionize synthesis and manufacturing, identify the next electrolyte to store and transport energy to protect the grid, and more. But these breakthrough discoveries require data.
Data to train next-generation AI models and interatomic potentials. Data to push the boundaries of what's computationally possible in molecular chemistry and lead the world in AI for science. Data that captures the full complexity of chemical systems, from small organic molecules to massive biomolecular complexes.
The OMol25 Electronic Structures dataset includes the raw DFT outputs, electronic densities, wavefunctions, and molecular orbital information for over 4M million high-accuracy quantum chemical calculations. We see this as a transformative opportunity to develop higher quality partial charges, partial spins, and advanced electronic features to unlock the next generation of physics-informed ML models.
The Materials Data Facility is proud to make these data available via the Eagle cluster at ALCF through a high-performance Globus endpoint. Given the dataset's unprecedented scale, we're first releasing all output data for a 4M random OMol25 split, with the full multi-petabyte dataset following based on community engagement.
For this first release, the data are quite raw, and as-created by the Meta team. There's a significant opportunity for the community to build tools that simplify access to these data, allow data query and browsing, create databases of calculated properties and descriptors, and much more. We intend to work on these topics with all of you.
We can't wait to see what you can do with these data!
Access Details: github.com/facebookresearch/…
Eagle was pioneered as the Petrel project, a new way to provide researchers access to high-quality, high-volume data by Ian Foster, Rachana Ananthakrishnan, Kyle Chard, Michael Papka, Rick Stevens, and others. Globus.org provides core platform capabilities (auth, data transfer, workflow automation, and compute) to over 600k researchers.
Thanks to support from NIST and James Warren for making the MDF vision of vast troves of open data to fuel discovery possible.
@mshuaibii, @zackulissi , @argonne, @argonne_lcf
Sam Blau retweeted
The @AIatMeta FAIR Chemistry team is excited to release the OMol25 Electronic Structures Dataset - 500 TB of electronic molecular data for the community in partnership with @BenBlaiszik's team at @argonne and @UChicago!
Details: fair-chem.github.io/molecule…
The OMol25 leaderboard is live! Our comprehensive evals seek to 1) guide MLIP development + 2) help users understand where models are reliable. Excited to see community engagement/competition reduce errors & yield better models that accelerate scientific discovery for everyone!
Excited to present the FAIR Chemistry Leaderboard - a centralized space for our team’s community benchmark efforts. We’re kicking things off today with the OMol25 leaderboard!
📊Leaderboard: huggingface.co/spaces/facebo…
🖥️Code: github.com/facebookresearch/…
Note that some numbers look different from what is reported in the paper for two reasons: 1) We found & removed test/eval cases with problematic DFT. 2) We switched from Sella to LBFGS for optimization to reduce the GPU time required to run the evals. All trends remained the same
Sam Blau retweeted
200% faster inference for molecular simulation — meet OrbMol.
OrbMol is built specifically for molecules, from metal complexes to biomolecules and electrolytes.
It delivers state-of-the-art accuracy, including protein-ligand binding energies, experimental accuracy in molecular dynamics simulations, and greater memory efficiency than the next best alternative - all open-source under Apache 2.
In this piece, we showcase OrbMol through the example of enzyme simulation, highlighting how it captures subtle interactions that traditional models often miss.
