Structure-Centric Machine Learning is an independent research program led by Ahmed Elmahdi, PhD. The program is open to licensing partnerships, research collaborations, and institutional discussions.
Inventor of the Structure-Centric Machine Learning paradigm and the family of foundational tools it has produced. PhD in Control Systems Engineering from Purdue University, with a specialization in networked control systems, optimization algorithms, and system stability analysis.
Seven years of applied engineering experience across industrial automation, building systems, and control engineering at Johnson Controls and Panasonic — experience that grounds the structure-centric paradigm in operational realities, not academic abstraction. The paradigm was developed independently as an extension of mathematical work on dimensionality and scale invariance.
The full family of inventions — the Structure-Centric Paradigm itself, AdaBox, AdaGraph, SCOPE, Graph-SCOPE, SLCD, and the Density-Aware Sampler — was conceived, developed, and validated by the principal investigator. Cross-domain validation (genomics, text, materials science) is conducted through collaboration with domain specialists, including a cancer genomics laboratory for the lung adenocarcinoma module-discovery work.
Companies whose product engine relies on clustering quality — social listening, trend detection, B2B intent data, customer segmentation — can license AdaGraph + SLCD as a drop-in replacement for their HDBSCAN step. Favorable terms for the first 1–2 reference customers. Annual licensing structure with reference rights.
Genomics tool companies (QIAGEN, Partek, 10x Genomics, Genedata, Seven Bridges, DNAnexus) and materials informatics platforms (Citrine, Materials Project) can integrate the high-D stack as a native clustering option. Significantly higher licensing tier reflects the downstream value of the discoveries the technology enables.
A research lab focused on the structure-centric paradigm, with the principal investigator as founding director, would systematically explore the seven-domain roadmap. Active discussions welcomed with international AI institutes — KAUST, MBZUAI, NTU Singapore, University of Luxembourg, Technical University of Munich.
Researchers in genomics, neuroscience, materials science, climate science, astronomy, and epidemiology with a discovery problem that fits the structure-centric paradigm: please reach out. The research roadmap is a live document, and domain partnerships shape its priorities.
For licensing inquiries, research collaboration proposals, or institutional partnership discussions:
ahmed@structurecentricml.com
For technical inquiries about the algorithms, please reference the arXiv preprint (arXiv:2603.13339) or any specific result from the validation pages of this site. The full benchmark notebooks are available under appropriate research collaboration agreements.