Benchmark-driven evaluation of time series foundation models and edge LLM inference systems at the University of Cyprus
3
papers here
3
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—
No output data yet: the profile lacks publication counts, venue records, or totals beyond the three reviewed preprints. This axis would measure cumulative paper output and its distribution over time and across venues once a fuller record is linked.
Impact—
Impact data is absent: no citation counts, h-index, or usage signals were provided for any of the three 2026 preprints. This axis would measure how often the work is cited and taken up by other researchers once citation records accumulate.
Focus0/100
Radar focus scores 0 because the field distribution is a flat three-way split: time series forecasting, AI agent benchmarking for edge LLM systems, and edge/cloud computer systems each hold one third of the three catalog papers. All three appeared in 2026, so the spread reads as early-stage breadth across benchmarking problems rather than scattered drift; each paper still shares a common evaluation-and-measurement methodology.
Collaboration44/100
Collaboration scores 44 from 8 distinct coauthors across the three 2026 preprints. Recurring partners include Demetris Trihinas, George Pallis, Panagiotis Michael, Demetris Paschalides, Marios D. Dikaiakos and Maysam Khatib, a predominantly University of Cyprus cluster. On every paper the researcher sits in the middle author position and is never corresponding author, so the moderate score matches a contributor role inside an established local group rather than independent team leadership.
Activity—
Activity cannot be scored: only three papers, all dated 2026, are in the catalog, giving no multi-year cadence. This axis would measure annual publication counts and publishing consistency over recent years once a longer record is available.
Reproducibility70/100
Reproducibility is the strongest scored axis: Evolve computes a mean of 70 across the three reviewed papers. The zero-shot time series foundation model benchmark is rated high readiness, while MCP-GRANITE and the edge continuum measurement study are each medium, with Evolve flagging fixed-response mocks, confounded quantization, thin statistics, and non-comparable energy boundaries. A 70 mean reflects solid methodological care with room to tighten statistical reporting.
Research summary
Moysis Symeonides is a researcher at the University of Cyprus whose recent work concentrates on rigorous benchmarking and measurement of emerging AI systems. The Evolve catalog lists three arXiv preprints, all from 2026, in which Symeonides appears as a middle author rather than corresponding author. The output splits evenly — one third each — across time series forecasting, AI agent benchmarking for edge LLM systems, and edge/cloud computer systems, a three-way spread that produces a focus score of 0 and signals deliberate breadth rather than a single niche. One paper benchmarks six zero-shot time series foundation models on three datasets, identifying an accuracy–calibration trade-off; another, MCP-GRANITE, tests how tool-interface granularity affects MCP-based LLM agents on edge hardware; the third measures LLM inference trade-offs across edge continuum hardware, including energy measurement on Jetson AGX Orin. Evolve rates reproducibility mean 70 across these three papers (one high, two medium readiness), the strongest quantitative axis in the profile. Collaboration scores 44 with 8 distinct coauthors, largely University of Cyprus colleagues such as Demetris Trihinas, George Pallis and Marios D. Dikaiakos. Output, impact and activity metrics are not yet available.
Collaboration pattern
All three 2026 preprints place Symeonides as a middle author with 8 distinct coauthors, yielding a moderate collaboration score of 44. Recurring partners — Demetris Trihinas, George Pallis, Panagiotis Michael, Demetris Paschalides, Marios D. Dikaiakos and Maysam Khatib — form a University of Cyprus systems-and-ML cluster, suggesting benchmarking expertise contributed within an established local team rather than corresponding-author leadership.
Research keywords
Highlights
Zero-shot time series foundation models show an accuracy–calibration trade-off
MCP-GRANITE: a four-tool interface best aids edge LLM agents
Measuring LLM inference trade-offs across edge continuum hardware
Timeline
- 2026Benchmarked six zero-shot time series foundation models across three datasets, finding an accuracy–calibration trade-off.
- 2026Published MCP-GRANITE, showing a 4-tool interface granularity best aids MCP-based LLM agents at the edge.
- 2026Measured LLM inference trade-offs across edge continuum hardware, including energy on Jetson AGX Orin.
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 only three 2026 arXiv preprints reviewed in the Evolve catalog, not a complete publication record.
- Output, impact and activity axes lack underlying data (no citation counts or multi-year output were provided) and are therefore unscored.
- All catalog papers are arXiv preprints with middle-author placement and no corresponding-author role; no peer-reviewed venue information is available.
- Identity is name-resolved from public bibliographic data via ORCID 0009-0007-2711-1949; the supplied coauthor list contains the researcher's own name and a duplicate, so counts follow the raw metrics as given.