Advancing materials discovery and design through AI, physics-based simulation, and experimental integration
Our Mission
Materials are everywhere! From the semiconductors powering your devices to the superconductors enabling quantum computers, materials define what’s technologically possible. With an estimated 10100 possible materials, discovering the right one is an immense challenge. We’re building AI agents to make this search 100 to 1000× faster.
What We Work On
🔬 Target Materials
- High-temperature superconductors
- Next-generation semiconductors
- Advanced solar cells
- Dielectrics & piezoelectrics
- 2D materials & heterostructures
🤖 AI Methods
- Graph Neural Networks (GNNs)
- Generative Pre-trained Transformers
- Vision-Language Models
- Agentic AI frameworks
- Large Language Models for science
⚛️ Physics-Based Simulation
- Density Functional Theory (DFT)
- Tight-binding models
- Molecular dynamics
- Quantum Monte Carlo
- Finite element modeling
The JARVIS Ecosystem
We’ve built the Joint Automated Repository for Various Integrated Simulations (JARVIS), a comprehensive infrastructure hosting over 1 million material properties. This massive database serves as the training ground for our AI models, which learn to predict and generate new materials with desired properties. Our research spans the Processing-Structure-Property-Performance (PSPP) paradigm across multiple length and time scales.
Multi-Modal Data Integration
We work with diverse data types to enable comprehensive materials understanding:
- Scalars: Formation energies, bandgaps, elastic constants
- Spectra: X-ray diffraction (XRD), EELS
- Images: STEM and STM microscopy
- Text: Scientific literature, documentation
Open Science Commitment
All our research outputs are freely available to accelerate global materials discovery:
Collaborative Research
We actively collaborate with experimental groups, national laboratories, and industry partners to validate our computational predictions and accelerate the materials discovery pipeline. Our interdisciplinary approach combines expertise from materials science, physics, chemistry, computer science, and data science.