Klinikum Klagenfurt researcher in governed LLM decision support and efficient recurrent language-model architectures
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
0
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-15) 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—
This axis would measure publication volume and consistency over time. It is not scored: the snapshot shows only six recent works, all dated 2026, plus two cataloged arXiv papers, too thin a record to rate.
Impact—
This axis would measure citation-based influence, such as per-paper citations and field-weighted reach. It is not scored: the snapshot lists h-index and i10-index as 0 with no citation aggregates, and a 2026-only record has had little time to accrue citations.
Focus0/100
The focus score is 0 because the record splits evenly across two distinct fields: AI governance / trustworthy LLM systems and machine learning for efficient language models, each weighted 0.5. Topic counts confirm the spread: topic modeling, NLP techniques, ethics of AI, and adversarial robustness each appear twice, with neural-network applications once. This is a dual-track profile rather than a narrow specialization, which the metric reflects.
Collaboration36/100
The collaboration score is 36, reflecting a small working circle: five distinct co-authors in the snapshot, with Kabeh Mohsenzadegan and Kyandoghere Kyamakya recurring most often. Both cataloged papers list him in a middle author position without corresponding authorship, suggesting contribution within a team led by others rather than an independent group. The frequent-coauthor list also contains duplicate entries and his own name, indicating sparse, partially noisy bibliographic metadata.
Activity57/100
Activity scores 57, on the moderate side. The record shows six works in the last three years, with the counts-by-year data placing all six in 2026, and both Evolve-cataloged arXiv papers are dated 2026. This suggests a recent burst of output rather than sustained multi-year production; earlier years show no indexed works in this snapshot, so the trajectory is too short to judge momentum.
Reproducibility78/100
Reproducibility scores 78, computed from two Evolve-assessed papers (mean 78 across two repro-scored papers). Policy-as-Skill is rated high readiness, as governance packaged as versioned executable skills with evidence and audit naturally aids replication, while TinyCeNN-LM is medium, with honest but small-scale evidence. Both verdicts flag limitations, yet the transparency of stated constraints supports above-average reproducibility for an early record.
Research summary
Vahid Tavakkoli is a researcher at Klinikum Klagenfurt in Austria whose profile is name-resolved from public OpenAlex bibliographic data, without a linked ORCID. The visible record is recent and concentrated: six works in the last three years, all counted in 2026, including two arXiv papers reviewed in the Evolve catalog. His fields split evenly (0.5 and 0.5) between AI governance for trustworthy LLM systems and machine learning for efficient language models, a dual track that yields a focus score of 0. The first paper, Policy-as-Skill, packages governed LLM decision support as versioned executable skills with evidence grounding, deterministic control, and audit trails; Evolve rates its reproducibility readiness high while noting that headline gains ride a degenerate subset, a synthetic corpus, and single-expert labels. The second, TinyCeNN-LM, converts pretrained attention into CeNN-inspired cellular-recurrent layers under a quality gate with accept-or-rollback; the evidence is honest but small-scale, with no speedup and marginal cache gains. Mean reproducibility across the two assessed papers is 78. Activity is moderate (57), collaboration modest (36) with five distinct co-authors, most often Kabeh Mohsenzadegan and Kyandoghere Kyamakya, in middle author positions without corresponding authorship. Citation impact is not yet scorable: the snapshot lists h-index and i10-index of 0.
Collaboration pattern
Collaboration is small-scale: five distinct co-authors, with Kabeh Mohsenzadegan and Kyandoghere Kyamakya the most frequent partners in the record. Both cataloged papers place him in a middle author position without corresponding authorship, consistent with contributing inside a small team. The frequent-coauthor list contains duplicates and his own name, so partner metadata should be read with care. The collaboration dimension scores 36.
Research keywords
Highlights
Policy-as-Skill: governance as versioned executable skills for LLM decision support
TinyCeNN-LM: quality-gated attention-to-recurrent conversion
Timeline
- 2026Published 'Policy-as-Skill: Governed LLM Decision Support with Evidence, Deterministic Control, and Audit' on arXiv; Evolve rates its reproducibility readiness as high.
- 2026Published 'TinyCeNN-LM: Quality-Gated Conversion of Pretrained Attention with CeNN-Inspired Cellular-Recurrent Layers' on arXiv, proposing attention replacement with accept-or-rollback.
- 2026OpenAlex counts six works for this year, comprising the entire visible three-year record; both cataloged papers list him as a middle author.
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 figures come from an OpenAlex snapshot only; no ORCID is linked, so the profile is name-resolved from public bibliographic data and may merge or miss records.
- The output and impact dimensions are unscored: the snapshot lists h-index and i10-index of 0 with no citation aggregates.
- The reproducibility score (78) and all radar dimensions rest on only two Evolve-reviewed papers, both from 2026; the record is too short for stable trends.
- The frequent-coauthor list contains duplicate entries and the researcher's own name; only two external partners were reliably identifiable.