Sovereign AI infrastructure
How institutions can use AI while keeping data residency, control and accountability inside their own boundary.
AdiOS is built in Rust as an operating system for institutions that want to adopt AI and keep it sovereign. This page collects the technical papers, preprints and author profiles behind it.
Founder, AdiOS Platform Private Limited
Malay Baral has spent about thirty years working in data and AI. He was Chief Technology Officer of T-Mobile Netherlands and has held chief architect roles at IBM, Capgemini, CA Technologies, EPAM and SBI. He has also designed public-sector data systems for the European Union.
He founded AdiOS in 2026 to build a sovereign AI operating system: infrastructure where an institution's data, and the intelligence built on it, stay inside its own walls. His research writing sets out the problem, the hypothesis and the architecture behind that design.
The questions the AdiOS paper series works through.
How institutions can use AI while keeping data residency, control and accountability inside their own boundary.
Hexagonal, plugin and microkernel designs, and what they offer for isolation, extensibility and assurance.
Semantic and graph representations of institutional knowledge, and how they sit alongside AI workloads.
A multi-part series, published as technical preprints. Each paper carries a DOI once posted.
The opening paper of the series. It defines the problem AdiOS addresses, states five requirements for a sovereign AI operating system, and proposes a central hypothesis with four falsifiable sub-claims.
Version DOI: 10.5281/zenodo.23800938 All versions: 10.5281/zenodo.23800937 PDF
Scholarly identities for Malay Baral. Links appear here as each profile goes live.
Please cite Part I by its DOI. Use the all-versions DOI (10.5281/zenodo.23800937) to refer to the paper regardless of version.
Baral, M. (2026). The Circular Operating System, Part I: Problem Statement and Hypothesis for a Sovereign AI Operating System. Zenodo. https://doi.org/10.5281/zenodo.23800938
Questions about the research, or interested in reviewing a draft?
Write to the author