Agentic AI architecture
A conceptual model connects LLM reasoning, agent functions, feedback and external interfaces.
Gill et al. (2026), Agentic AI: Vision and challenges, PLOS Complex Systems, Fig. 1. CC BY 4.0. Unmodified. · CC BY 4.0
SourceResearch figures and photographs used in this portfolio.
The portrait is from my existing academic website.
A conceptual model connects LLM reasoning, agent functions, feedback and external interfaces.
Gill et al. (2026), Agentic AI: Vision and challenges, PLOS Complex Systems, Fig. 1. CC BY 4.0. Unmodified. · CC BY 4.0
SourceA domain map organises proposed applications across six sectors.
Gill et al. (2026), Agentic AI: Vision and challenges, PLOS Complex Systems, Fig. 2. CC BY 4.0. Unmodified. · CC BY 4.0
SourceA conceptual research map groups technical, trust, learning, infrastructure, multi-agent and governance themes.
Gill et al. (2026), Agentic AI: Vision and challenges, PLOS Complex Systems, Fig. 3. CC BY 4.0. Unmodified. · CC BY 4.0
SourceA layered conceptual comparison places processing capabilities across cloud, fog and edge infrastructure.
Gill et al. (2024), Edge AI: A Taxonomy, Systematic Review and Future Directions, arXiv:2407.04053v2, Fig. 1. CC BY 4.0. Unmodified. · CC BY 4.0 (author-deposited arXiv v2)
SourceA conceptual architecture shows data from IoT devices flowing to edge nodes, with AI results returning.
Gill et al. (2024), Edge AI: A Taxonomy, Systematic Review and Future Directions, arXiv:2407.04053v2, Fig. 2. CC BY 4.0. Unmodified. · CC BY 4.0 (author-deposited arXiv v2)
SourceA domain map links six edge-AI application areas with their processing objectives.
Gill et al. (2024), Edge AI: A Taxonomy, Systematic Review and Future Directions, arXiv:2407.04053v2, Fig. 3. CC BY 4.0. Unmodified. · CC BY 4.0 (author-deposited arXiv v2)
SourceThe system diagram links workload analysis and DeepAR forecasting to resource management and VM autoscaling.
Velu, Gill, Murugesan, Wu and Li, CloudAIBus (2024), author repository. Apache-2.0 repository licence. Unmodified. · Apache-2.0 at repository level; no separate image-specific notice found
SourceConceptual overview of RF capsule localisation: wearable anchors, a cylindrical model and a learned conditional flow.
Murugesan et al., EuCAP 2026 paper source. · Author-supplied research figure, included in this private draft.
SourceThe large intestine. Anatomical context for research into gastrointestinal shape variation.
National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health. · NIDDK media-library image, reproduced unchanged with requested credit.
SourceA quadruped robot and wheeled Jackal platform in the demonstration area at the EME Hub Summer School, 31 July 2026.
Photograph supplied by Subramaniam Murugesan. · Used with the portfolio owner’s permission. Source capture metadata removed; photograph unchanged.
SourceThe University of Glasgow during my visit for the EME Hub Summer School, July 2026.
Photograph supplied by Subramaniam Murugesan. · Used with the portfolio owner’s permission. Source capture metadata removed; photograph unchanged.
SourceServerless deployment architecture for tuberculosis prediction using Amazon EKS and AWS Fargate.
Fig. 3 from S. S. Murugesan, S. Velu, M. Golec, H. Wu and S. S. Gill, IEEE Journal of Biomedical and Health Informatics 28(9), 5043–5054 (2024), doi:10.1109/JBHI.2024.3367736. © 2024 IEEE. · Not openly licensed. IEEE copyright with author-retained reuse provisions.
SourceGAIKube combines generative workload data, TimesFM forecasting and Kubernetes container orchestration.
GAIKube framework by Babar Ali et al., author repository accompanying doi:10.1109/TCCN.2024.3508771. © 2024 Babar Ali. BSD-3-Clause repository licence. · BSD-3-Clause at repository/documentation level; no separate figure-specific licence notice found.
SourceCE-Track at a MedTech Venture Builder event, July 2026. A frame from the video supplied for this portfolio.
Video supplied by Subramaniam Murugesan. · Used with the portfolio owner’s permission. Original composition retained; source capture metadata removed.
SourceDinner with fellow attendees at EuCAP 2026 in Dublin.
Photograph supplied by Subramaniam Murugesan. · Used with the portfolio owner’s permission. Source metadata removed; primary photograph unchanged.
SourceDublin waterfront during the EuCAP 2026 visit.
Photograph supplied by Subramaniam Murugesan. · Used with the portfolio owner’s permission. Source metadata removed; primary photograph unchanged.
SourceAttending a conference talk at EuCAP 2026.
Photograph supplied by Subramaniam Murugesan. Presentation content belongs to its authors. · Used with the portfolio owner’s permission. Source metadata removed; primary photograph unchanged.
SourceAn evening in Dublin during EuCAP 2026.
Photograph supplied by Subramaniam Murugesan. · Used with the portfolio owner’s permission. Source metadata removed; primary photograph unchanged.
SourceAttending a EuCAP 2026 session on tissue-mimicking phantoms for antenna and propagation modelling.
Photograph supplied by Subramaniam Murugesan. Presentation content belongs to its authors. · Used with the portfolio owner’s permission. Source metadata removed; primary photograph unchanged.
Source