Kebotix
Accelerating materials discovery through the world's first self-driving laboratory ecosystem.

The industry-standard open-source library for data mining and feature engineering in materials science.
Matminer is a sophisticated Python-based toolkit designed to bridge the gap between materials science data and machine learning. As of 2026, it serves as the foundational architecture for Materials Informatics (MI), providing high-level abstractions for converting raw structural and compositional data into machine-learnable numerical descriptors, a process known as featurization. Its technical architecture is built atop the Pymatgen ecosystem, enabling seamless manipulation of crystallographic information files (CIF) and structural objects. Matminer’s core strength lies in its extensive library of over 70 featurizers, ranging from simple stoichiometric descriptors to complex graph-based and Voronoi-tessellation-derived structural features. In the 2026 market, Matminer is positioned as the essential pre-processing layer for autonomous laboratories and generative AI models in material discovery. By standardizing the interface for data retrieval from major databases like the Materials Project and Citrination, it enables reproducible research and rapid prototyping of predictive models for properties such as thermal conductivity, band gaps, and catalytic activity. It is widely adopted by both academic institutions and R&D departments in the battery, semiconductor, and aerospace industries to accelerate the 'Materials Genome' initiative.
Includes over 70 distinct featurizers covering composition, site-specific, and structural properties.
Accelerating materials discovery through the world's first self-driving laboratory ecosystem.
The OS for AI-driven materials discovery and manufacturing optimization.
Accelerate materials discovery with a unified Materials Informatics platform and AI-driven R&D.
Accelerating materials discovery through physics-aware generative AI and inverse design.
Verified feedback from the global deployment network.
Post queries, share implementation strategies, and help other users.
Unified wrappers for Materials Project, Citrination, Figshare, and MPDS.
Direct integration with Pandas for parallelized featurization across multiple CPU cores.
Calculates structural features based on the geometry of Voronoi polyhedra surrounding atoms.
Implementation of the BoB method for molecular and crystal systems.
Built-in tools for versioning and caching large-scale materials datasets.
Direct access to the Materials Agnostic Platform for Informatics and Exploration (Magpie) descriptors.
Manual DFT calculations for band gaps are computationally expensive.
Registry Updated:2/7/2026
Identifying stable electrode materials from millions of possible chemistries.
Optimizing surface binding energy for specific gas molecules.