Ben Blaiszik
@benblaiszik.bsky.social
Group Leader - AI and data infrastructure for science at
UChicago/Argonne/Globus - UofIllinois alum. materials, chemistry, physics. Opinions are my own.🤖🔬
UChicago/Argonne/Globus - UofIllinois alum. materials, chemistry, physics. Opinions are my own.🤖🔬
So your data are available upon reasonable request? Well, we are making some reasonable requests - at scale. :)
1. Search literature (currently stubbed)
2. Enumerate papers, extract contacts
3. Send email w/ data drop location
4. Parse data
Does anyone want to help productionize this?
1. Search literature (currently stubbed)
2. Enumerate papers, extract contacts
3. Send email w/ data drop location
4. Parse data
Does anyone want to help productionize this?
November 21, 2025 at 9:32 PM
So your data are available upon reasonable request? Well, we are making some reasonable requests - at scale. :)
1. Search literature (currently stubbed)
2. Enumerate papers, extract contacts
3. Send email w/ data drop location
4. Parse data
Does anyone want to help productionize this?
1. Search literature (currently stubbed)
2. Enumerate papers, extract contacts
3. Send email w/ data drop location
4. Parse data
Does anyone want to help productionize this?
Yesterday I did a glow-up of the entire Materials Data Facility front end using Gemini 3 in less than 2 hours. A few things I learned followed by few snapshots of the updates.
What did I learn?
1. Gemini 3 feels clearly ahead in terms of front-end design. Less AI slop, and more focus.
What did I learn?
1. Gemini 3 feels clearly ahead in terms of front-end design. Less AI slop, and more focus.
November 19, 2025 at 5:48 PM
Yesterday I did a glow-up of the entire Materials Data Facility front end using Gemini 3 in less than 2 hours. A few things I learned followed by few snapshots of the updates.
What did I learn?
1. Gemini 3 feels clearly ahead in terms of front-end design. Less AI slop, and more focus.
What did I learn?
1. Gemini 3 feels clearly ahead in terms of front-end design. Less AI slop, and more focus.
Northern Lights over the western Chicago suburbs at 9:15. Clearly visible by eye, lasted about 10 mins. 🤩
November 12, 2025 at 3:37 AM
Northern Lights over the western Chicago suburbs at 9:15. Clearly visible by eye, lasted about 10 mins. 🤩
🔥We're excited to announce a major milestone for the machine-learned interatomic potential (MLIP) ecosystem: TorchSim is moving to community ownership and governance through a partnership with Radical AI and the open-source community! TorchSim is an atomistic simulation engine built for the AI era.
October 15, 2025 at 7:06 PM
🔥We're excited to announce a major milestone for the machine-learned interatomic potential (MLIP) ecosystem: TorchSim is moving to community ownership and governance through a partnership with Radical AI and the open-source community! TorchSim is an atomistic simulation engine built for the AI era.
Generative AI is changing how we discover materials, but without direction, it can quickly lose its way. Marcus Schwarting led an effort to show that using active learning as a guide helps to prioritize the best candidates in scientific discovery workflows for MOFs.
October 7, 2025 at 5:39 PM
Generative AI is changing how we discover materials, but without direction, it can quickly lose its way. Marcus Schwarting led an effort to show that using active learning as a guide helps to prioritize the best candidates in scientific discovery workflows for MOFs.
Are you looking for ~120~ examples of how LLMs can be used to advance research in materials science & chemistry with open code? We've made a quick search interface for you to explore projects from the recent hackathon.
See the projects here and come join our community: llmhackathon.github.io
See the projects here and come join our community: llmhackathon.github.io
October 2, 2025 at 7:03 PM
Are you looking for ~120~ examples of how LLMs can be used to advance research in materials science & chemistry with open code? We've made a quick search interface for you to explore projects from the recent hackathon.
See the projects here and come join our community: llmhackathon.github.io
See the projects here and come join our community: llmhackathon.github.io
🔥 Today we announce the Meta OMol25 Electronic Structures Dataset - 500 TB of molecular data in collaboration with the AI at Meta team.
Access Details: github.com/facebookrese...
Access Details: github.com/facebookrese...
September 24, 2025 at 4:41 PM
🔥 Today we announce the Meta OMol25 Electronic Structures Dataset - 500 TB of molecular data in collaboration with the AI at Meta team.
Access Details: github.com/facebookrese...
Access Details: github.com/facebookrese...
Over 1100 people around the world, virtual and in-person, decided to take a chance together, to imagine, and build - seeking ways to speed discovery and understanding in materials and chemistry with AI.
We'll share our findings soon, but from what I've heard already prepare to be amazed.
We'll share our findings soon, but from what I've heard already prepare to be amazed.
September 10, 2025 at 3:43 PM
Over 1100 people around the world, virtual and in-person, decided to take a chance together, to imagine, and build - seeking ways to speed discovery and understanding in materials and chemistry with AI.
We'll share our findings soon, but from what I've heard already prepare to be amazed.
We'll share our findings soon, but from what I've heard already prepare to be amazed.
Read the entire paper and learn about more of our findings here by the amazing team at @argonne @argonne_lcf @UChicago @globus @ianfoster @ENERGY @labsglobus
Read the paper: arxiv.org/abs/2508.18489
Read the paper: arxiv.org/abs/2508.18489
September 9, 2025 at 3:14 PM
Read the entire paper and learn about more of our findings here by the amazing team at @argonne @argonne_lcf @UChicago @globus @ianfoster @ENERGY @labsglobus
Read the paper: arxiv.org/abs/2508.18489
Read the paper: arxiv.org/abs/2508.18489
We tested MCP servers in scientific workflows including:
⚛️ Chemistry/Materials (MLIPs in Garden, gaps w/ Globus Compute)
🧬 Bioinformatics (phylogenetics across ALCF+NERSC)
📂 Filesystem monitoring (Icicle + Octopus)
Paper: arxiv.org/abs/2508.18489
⚛️ Chemistry/Materials (MLIPs in Garden, gaps w/ Globus Compute)
🧬 Bioinformatics (phylogenetics across ALCF+NERSC)
📂 Filesystem monitoring (Icicle + Octopus)
Paper: arxiv.org/abs/2508.18489
September 9, 2025 at 3:14 PM
We tested MCP servers in scientific workflows including:
⚛️ Chemistry/Materials (MLIPs in Garden, gaps w/ Globus Compute)
🧬 Bioinformatics (phylogenetics across ALCF+NERSC)
📂 Filesystem monitoring (Icicle + Octopus)
Paper: arxiv.org/abs/2508.18489
⚛️ Chemistry/Materials (MLIPs in Garden, gaps w/ Globus Compute)
🧬 Bioinformatics (phylogenetics across ALCF+NERSC)
📂 Filesystem monitoring (Icicle + Octopus)
Paper: arxiv.org/abs/2508.18489
Imagine researchers being able to:
+ Move data effortlessly across systems
+ Launch simulations on Exascale HPC systems
+ Run AI models via Garden & Galaxy
All via user or agent intent/language to make the next breakthroughs in energy, materials, and chemistry.
+ Move data effortlessly across systems
+ Launch simulations on Exascale HPC systems
+ Run AI models via Garden & Galaxy
All via user or agent intent/language to make the next breakthroughs in energy, materials, and chemistry.
September 9, 2025 at 3:14 PM
Imagine researchers being able to:
+ Move data effortlessly across systems
+ Launch simulations on Exascale HPC systems
+ Run AI models via Garden & Galaxy
All via user or agent intent/language to make the next breakthroughs in energy, materials, and chemistry.
+ Move data effortlessly across systems
+ Launch simulations on Exascale HPC systems
+ Run AI models via Garden & Galaxy
All via user or agent intent/language to make the next breakthroughs in energy, materials, and chemistry.
How are we building this?
Models are made available through TorchSim significantly boosting performance
🔗 github.com/Radical-AI/t...
On cloud: models are available through credits from @Modal (thanks @bernhardsson) so researchers can try MLIPs without spinning up infra.
Models are made available through TorchSim significantly boosting performance
🔗 github.com/Radical-AI/t...
On cloud: models are available through credits from @Modal (thanks @bernhardsson) so researchers can try MLIPs without spinning up infra.
August 20, 2025 at 7:27 PM
How are we building this?
Models are made available through TorchSim significantly boosting performance
🔗 github.com/Radical-AI/t...
On cloud: models are available through credits from @Modal (thanks @bernhardsson) so researchers can try MLIPs without spinning up infra.
Models are made available through TorchSim significantly boosting performance
🔗 github.com/Radical-AI/t...
On cloud: models are available through credits from @Modal (thanks @bernhardsson) so researchers can try MLIPs without spinning up infra.
This release includes access to MLIPs such as MACE, MatterSim,
SevenNet and more.
We’ve created initial benchmarks for task performance, cloud cost, and HPC compute cost for each model.
See the MLIP Garden here: thegardens.ai#/use-cases/m...
SevenNet and more.
We’ve created initial benchmarks for task performance, cloud cost, and HPC compute cost for each model.
See the MLIP Garden here: thegardens.ai#/use-cases/m...
August 20, 2025 at 7:27 PM
This release includes access to MLIPs such as MACE, MatterSim,
SevenNet and more.
We’ve created initial benchmarks for task performance, cloud cost, and HPC compute cost for each model.
See the MLIP Garden here: thegardens.ai#/use-cases/m...
SevenNet and more.
We’ve created initial benchmarks for task performance, cloud cost, and HPC compute cost for each model.
See the MLIP Garden here: thegardens.ai#/use-cases/m...
🚀 We asked: What if the time to set up and run the best Machine-Learned Interatomic Potentials (MLIPs) took seconds, not days?
Today, we release the MLIP Garden v0.1.
What you can do now:
- Experiment ~instantly
- Scale deployments on experimental NSF and Dept of Energy systems
Today, we release the MLIP Garden v0.1.
What you can do now:
- Experiment ~instantly
- Scale deployments on experimental NSF and Dept of Energy systems
August 20, 2025 at 7:27 PM
🚀 We asked: What if the time to set up and run the best Machine-Learned Interatomic Potentials (MLIPs) took seconds, not days?
Today, we release the MLIP Garden v0.1.
What you can do now:
- Experiment ~instantly
- Scale deployments on experimental NSF and Dept of Energy systems
Today, we release the MLIP Garden v0.1.
What you can do now:
- Experiment ~instantly
- Scale deployments on experimental NSF and Dept of Energy systems
BLASTNet is a much friendlier dataset than it sounds, and is especially useful for those working in space propulsion, chemical reaction optimization, fusion physics, aerospace applications, weather modeling, and more. The team is seeking to collect the highest quality datasets, and needs your help.
August 13, 2025 at 5:52 PM
BLASTNet is a much friendlier dataset than it sounds, and is especially useful for those working in space propulsion, chemical reaction optimization, fusion physics, aerospace applications, weather modeling, and more. The team is seeking to collect the highest quality datasets, and needs your help.
Today we are opening the teaming board for the LLM Hackathon for Applications in Materials and Chemistry (Sept 11-12).
This is your opportunity to rally a team to help solve big problems!
Register here for the link: llmhackathon.github.io
This is your opportunity to rally a team to help solve big problems!
Register here for the link: llmhackathon.github.io
August 5, 2025 at 7:55 PM
Today we are opening the teaming board for the LLM Hackathon for Applications in Materials and Chemistry (Sept 11-12).
This is your opportunity to rally a team to help solve big problems!
Register here for the link: llmhackathon.github.io
This is your opportunity to rally a team to help solve big problems!
Register here for the link: llmhackathon.github.io
You can grow fresh peaches in your yard as far north as Chicago. Quite the unlock.
July 25, 2025 at 7:24 PM
You can grow fresh peaches in your yard as far north as Chicago. Quite the unlock.
We're about a month and a half away from the 3rd LLM Hackathon for Applications in Materials and Chemistry (Sept 11-12). I hope you will join us! Here are a few reasons to register now.
🤗 Join the preeminent community for LLM and multimodal model applications in materials science and chemistry.
🤗 Join the preeminent community for LLM and multimodal model applications in materials science and chemistry.
July 22, 2025 at 7:26 PM
We're about a month and a half away from the 3rd LLM Hackathon for Applications in Materials and Chemistry (Sept 11-12). I hope you will join us! Here are a few reasons to register now.
🤗 Join the preeminent community for LLM and multimodal model applications in materials science and chemistry.
🤗 Join the preeminent community for LLM and multimodal model applications in materials science and chemistry.
A key insight. Humans naturally "distort" objects mentally when comparing them - and this model is built to do the same.
Using this approach, they achieve near-human performance on character/doodle recognition without transfer learning and with one or few examples.
Using this approach, they achieve near-human performance on character/doodle recognition without transfer learning and with one or few examples.
July 21, 2025 at 5:45 PM
A key insight. Humans naturally "distort" objects mentally when comparing them - and this model is built to do the same.
Using this approach, they achieve near-human performance on character/doodle recognition without transfer learning and with one or few examples.
Using this approach, they achieve near-human performance on character/doodle recognition without transfer learning and with one or few examples.
🚀 Announcing the third LLM Hackathon for Applications in Materials and Chemistry (Sept 11-12)!
Last year, 34 teams submitted inspiring and open examples. This year, we are expecting amazing applications with higher powered models and agentic frameworks. Your imagination is the only limit.
Last year, 34 teams submitted inspiring and open examples. This year, we are expecting amazing applications with higher powered models and agentic frameworks. Your imagination is the only limit.
July 10, 2025 at 1:09 PM
🚀 Announcing the third LLM Hackathon for Applications in Materials and Chemistry (Sept 11-12)!
Last year, 34 teams submitted inspiring and open examples. This year, we are expecting amazing applications with higher powered models and agentic frameworks. Your imagination is the only limit.
Last year, 34 teams submitted inspiring and open examples. This year, we are expecting amazing applications with higher powered models and agentic frameworks. Your imagination is the only limit.
➡️ You can implement this approach with a single, reusable system prompt to deliver a deeper “chain-of-reasoning”.
Wonderful work by the team at Argonne, UIC, and NU including Hassan Harb, Rajeev Assary, Brian Ingram and others.
Paper link: doi.org/10.26434/che...
Prompt: github.com/HassanHarb92...
Wonderful work by the team at Argonne, UIC, and NU including Hassan Harb, Rajeev Assary, Brian Ingram and others.
Paper link: doi.org/10.26434/che...
Prompt: github.com/HassanHarb92...
July 8, 2025 at 8:25 PM
➡️ You can implement this approach with a single, reusable system prompt to deliver a deeper “chain-of-reasoning”.
Wonderful work by the team at Argonne, UIC, and NU including Hassan Harb, Rajeev Assary, Brian Ingram and others.
Paper link: doi.org/10.26434/che...
Prompt: github.com/HassanHarb92...
Wonderful work by the team at Argonne, UIC, and NU including Hassan Harb, Rajeev Assary, Brian Ingram and others.
Paper link: doi.org/10.26434/che...
Prompt: github.com/HassanHarb92...
Why does this matter?
➡️ The prompted LLM saw improved performance, achieving SOTA (97.15%) on the ARC Challenge without fine-tuning or external tool usage.
➡️ The approach showed consistent gains in reasoning depth, clarity, and domain-specific insight across chemistry and materials science.
➡️ The prompted LLM saw improved performance, achieving SOTA (97.15%) on the ARC Challenge without fine-tuning or external tool usage.
➡️ The approach showed consistent gains in reasoning depth, clarity, and domain-specific insight across chemistry and materials science.
July 8, 2025 at 8:25 PM
Why does this matter?
➡️ The prompted LLM saw improved performance, achieving SOTA (97.15%) on the ARC Challenge without fine-tuning or external tool usage.
➡️ The approach showed consistent gains in reasoning depth, clarity, and domain-specific insight across chemistry and materials science.
➡️ The prompted LLM saw improved performance, achieving SOTA (97.15%) on the ARC Challenge without fine-tuning or external tool usage.
➡️ The approach showed consistent gains in reasoning depth, clarity, and domain-specific insight across chemistry and materials science.
A new preprint introduces a Socratic Method Agent that guides LLMs through structured questioning, i.e., definition, analogy, hypothesis elimination, and more following lessons learned from thousands of years of philosophy and reasoning achieving SOTA on the ARC Challenge.
Continued below w links:
Continued below w links:
July 8, 2025 at 8:25 PM
A new preprint introduces a Socratic Method Agent that guides LLMs through structured questioning, i.e., definition, analogy, hypothesis elimination, and more following lessons learned from thousands of years of philosophy and reasoning achieving SOTA on the ARC Challenge.
Continued below w links:
Continued below w links:
Even more open materials and chemistry data for everyone! Think of all the models you could train using these datasets.
June updates: 26 high-quality datasets spanning polymer science, drug discovery, spectroscopy, MOF databases, + foundation model training datasets.
🔗: github.com/blaiszik/awe...
June updates: 26 high-quality datasets spanning polymer science, drug discovery, spectroscopy, MOF databases, + foundation model training datasets.
🔗: github.com/blaiszik/awe...
June 24, 2025 at 3:30 PM
Even more open materials and chemistry data for everyone! Think of all the models you could train using these datasets.
June updates: 26 high-quality datasets spanning polymer science, drug discovery, spectroscopy, MOF databases, + foundation model training datasets.
🔗: github.com/blaiszik/awe...
June updates: 26 high-quality datasets spanning polymer science, drug discovery, spectroscopy, MOF databases, + foundation model training datasets.
🔗: github.com/blaiszik/awe...
