Omran Berjawi
—
ML researcher spanning email threat detection, static malware detection, and LLM-based social simulation
3
papers here
3
reviewed by Evolve
2
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-14) 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—
Not yet scored: the underlying data is unavailable — total publication volume, venue breakdown, and output beyond the three Evolve-reviewed arXiv papers. This axis would measure research productivity and the scale of the researcher's published output.
Impact—
Not yet scored: no citation counts, h-index, or downstream uptake are present in the input data. This axis would measure scholarly influence — how often the work is cited, reused, and taken up by other researchers over time.
Focus0/100
Focus scores 0, matching the raw field distribution: the three reviewed papers split evenly at about 0.333 each across email threat detection, malware detection, and LLM-based social simulation. The portfolio is broad rather than concentrated, pairing two security applications with one computational social science direction.
Collaboration48/100
Collaboration scores 48 on a base of 10 distinct coauthors across the three reviewed papers, pointing to small working teams rather than large consortia. The frequent-coauthor list includes Rida Khatoun, Walid Fahs, Giuseppe Fenza, and Sherali Zeadally, and two papers carry Berjawi as first author, suggesting a lead-contributor role within compact groups.
Activity—
Not yet scored: only three papers, all dated 2026, are available, so no multi-year publication cadence can be established. This axis would measure research pace and continuity of output over successive years.
Reproducibility55/100
Reproducibility scores 55, the mean readiness across all three reviewed papers, each rated medium. Evolve's verdicts flag concrete gaps this axis tracks: AURA lacks a no-routing baseline, the opinion-dynamics work released no code, and Delphi Scanner shows dataset accounting inconsistencies — leaving clear, identifiable room for improvement.
Research summary
Omran Berjawi is an early-career machine learning researcher whose Evolve-cataloged output — three arXiv papers dated 2026 — spans two strands of applied AI: machine learning for cybersecurity and LLM-based computational social science. As first author, Berjawi proposed AURA, an adaptive uncertainty-routed email threat detection framework combining multimodal fusion and LoRA fine-tuning, reporting out-of-distribution F1 scores of 0.944–0.950; Evolve's review notes the routing gain lacks a no-routing baseline. In a second first-author paper, a generative-agent framework builds LLM digital twins of social network users for opinion dynamics, claiming more than 50% lower MAE than classical models on two Twitter datasets — promising, per Evolve, but evaluated on small networks with sentiment-proxy ground truth and no released code. A third paper, Delphi Scanner, lists Berjawi in a middle author position and applies CNN modeling to Windows PE API call sequences for static malware detection, reporting 95.35% accuracy, a 1.53MB model, and sub-millisecond inference. Across the three reviewed papers, Evolve computes a mean reproducibility readiness of 55 (all rated medium), ten distinct coauthors, and an even three-way split of topic focus; citation impact, total output volume, and multi-year activity cannot yet be scored.
Collaboration pattern
The reviewed output comes from small teams: 10 distinct coauthors across three papers, with Berjawi first author on two and in a middle position on one. The frequent-coauthor list is anchored by Rida Khatoun and Walid Fahs, and also includes Bijied Brahimi, Vincent Cohadon, Gabriel Glazman, Rayan Al Mohaize, Giuseppe Fenza, and Sherali Zeadally — a mix suggesting compact security-oriented groups plus at least one collaboration bridging into computational social science.
Research keywords
Highlights
Uncertainty-routed email threat detection with strong OOD F1
LLM digital twins for opinion dynamics simulation
Compact CNN for static Windows malware detection
Timeline
- 2026First-author AURA paper on adaptive uncertainty-routed email threat detection posted to arXiv.
- 2026First-author framework for generative LLM digital twins of opinion dynamics posted to arXiv.
- 2026Co-authored Delphi Scanner on efficient, interpretable static malware detection via API sequence modeling.
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 signals derive from Evolve's review of only three arXiv papers (all 2026); no citation counts, h-index, or venue history are included.
- Institution, country, and ORCID are absent from the input data; the profile is name-resolved from public bibliographic records and has not been ORCID-verified.
- Output, impact, and activity dimensions are unscored because the underlying metrics are unavailable, not because they are zero.
- Author-position and coauthor information come from the Evolve catalog snapshot and may not reflect the researcher's full publication record.