University of Cyprus systems researcher benchmarking LLM inference and MCP agent interfaces across the edge continuum
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
Output metrics (publication volume, venue spread, and authorship roles over time) are not yet computed for this profile, so no score is shown; the axis would measure how much this researcher publishes and in what venues.
Impact—
Impact metrics such as citation counts, h-index, or downstream uptake are not available in this snapshot; the axis would measure how widely this researcher's work is cited and adopted, and is left unscored here.
Focus0/100
The focus score is 0, derived from an even 0.5/0.5 field distribution across two adjacent areas: distributed systems / AI agent benchmarking (edge LLM systems) and computer systems (edge/cloud computing). The two 2026 cataloged papers sit on either side of this split — one on MCP agent tool interfaces, one on LLM inference measurement on edge hardware. The low score reflects balanced coverage of two tightly related edge-systems topics rather than scattered interests.
Collaboration41/100
The collaboration dimension scores 41, based on 7 distinct coauthors in the bibliometric snapshot. The recurring collaborator list is dominated by a University of Cyprus cluster — Moysis Symeonides, Demetris Paschalides, and Marios D. Dikaiakos appear repeatedly, with Maysam Khatib and Demetris Trihinas also frequent. Both 2026 papers carry middle, non-corresponding authorship, suggesting participation in an established local group rather than an independent or internationally distributed network so far.
Activity—
Activity metrics — publication cadence, recency of output, and yearly throughput — are not yet computed from the available data; the axis would measure how consistently and recently this researcher publishes, and currently carries no score.
Reproducibility55/100
Reproducibility scores a mean of 55 across the 2 papers assessed, both rated medium readiness by Evolve's review. The MCP-GRANITE benchmark is reproducible in structure but leans on fixed-response mocks and a quantization setup that confounds its interface comparison; the edge-hardware measurement study is carefully hedged yet has thin statistics and non-comparable energy boundaries. The mid-range score reflects solid, documented methodology tempered by these evidence limitations.
Research summary
George Pallis is a researcher at the University of Cyprus whose recent work centers on benchmarking large language model (LLM) systems at the edge. Two 2026 arXiv papers in the Evolve catalog define his current footprint. MCP-GRANITE tests tool-interface granularity for MCP-based LLM agents; Evolve's review credits its finding that a four-tool interface best aids edge agents, while cautioning that evidence rests on fixed-response mocks and confounded quantization. The second paper is a measurement study of LLM inference trade-offs across edge continuum hardware, rated careful and honestly hedged with solid methodology, though statistics are thin and energy boundaries are not comparable. Bibliometrically, the snapshot splits his output evenly across two adjacent fields — distributed systems / AI agent benchmarking (edge LLM systems) and computer systems (edge/cloud computing), a 0.5/0.5 distribution yielding a focus score of 0. Collaboration scores 41, with 7 distinct coauthors and a recurring Cyprus cluster around Moysis Symeonides, Demetris Paschalides, Marios D. Dikaiakos, Maysam Khatib, and Demetris Trihinas. Reproducibility averages 55 over the 2 papers assessed, both rated medium readiness. Output, impact, and activity metrics are unavailable, so this remains an early-stage profile built from a limited snapshot.
Collaboration pattern
Collaboration is concentrated in a University of Cyprus cluster: 7 distinct coauthors appear in the snapshot, with Moysis Symeonides, Demetris Paschalides, and Marios D. Dikaiakos recurring most often, joined by Maysam Khatib and Demetris Trihinas. Both 2026 cataloged papers carry middle, non-corresponding authorship, consistent with contributing within an established local systems group; no partners outside this cluster are visible in the available data.
Research keywords
Highlights
MCP-GRANITE: four-tool interfaces work best for edge LLM agents
Measuring LLM inference trade-offs across edge continuum hardware
Timeline
- 2026MCP-GRANITE benchmark for granularity interface testing of MCP-based LLM agents appears on arXiv, with middle authorship.
- 2026A measurement study of LLM inference trade-offs across edge continuum hardware is released on arXiv, also with middle authorship.
- 2026Both cataloged works land in the same year, concentrating this early-stage profile on edge LLM systems benchmarking within the University of Cyprus group.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Bibliometric scores are computed from OpenAlex-based public bibliographic data only; citation counts, h-index, and venue-level impact data are not included in this snapshot.
- Output, impact, and activity dimensions are missing from the radar because the underlying data was insufficient; no scores were estimated for them.
- The paper-level evidence base is two 2026 arXiv preprints reviewed by Evolve; peer-reviewed versions and earlier work are not covered.
- Identity is name-resolved via the provided ORCID (0000-0003-1815-5468) and institution; no country was supplied in the input data.