Catalyst Zero Research builds AI-native systems for materials discovery: knowledge graphs, scientific ML, structure/property reasoning, and research interfaces that reduce the time between a messy materials question and a grounded candidate.
The goal is simple: make scientific search feel less like days of isolated PDF digging and more like working inside a connected, inspectable research graph.
| Project | What it is | Links |
|---|---|---|
| Catalyst | AI-native materials discovery workspace with graph-grounded evidence, local candidate screening, 3D structure inspection, and an agentic research interface. | Live demo |
| TRIADS | Recursive neural architecture work for materials-property prediction and related materials ML workflows. | Repo |
- Materials-science knowledge graphs
- Agentic scientific research workflows
- Materials-property prediction
- Structure, spectra, thermodynamic, and electronic-property reasoning
- Human-facing tools that make scientific evidence easier to inspect and reuse
Catalyst lets a user ask for a material, screen candidates from a local Materials Project snapshot, open graph neighborhoods, inspect evidence, compare materials, and ask an agent to drive the workspace.
Live demo: catalyst-zero-research.github.io/Catalyst
Founder/operator: Rudra Tiwari
Portfolio: rtx09x.github.io