AI security and LLM benchmarking researcher — agent incident reporting and network protocol reasoning evaluation
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
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—
Not scored: this snapshot has no publication counts or venue volume for her. The axis would measure total research output — papers per year and their spread across venues — and needs a fuller bibliographic record than the two catalog entries provided.
Impact—
Not scored: no citation counts, h-index, or downstream-usage signals appear in this snapshot. The axis would measure how widely the work is cited and taken up — citation totals, citing venues, and influence on later systems — none of which the current data captures.
Focus0/100
Scored 0 on focus because the field distribution is an exact 50/50 split: 0.5 natural language processing (LLM benchmarking for network protocol analysis) and 0.5 AI security and governance. In this two-paper snapshot the score reflects deliberate dual-track positioning — one benchmarking study of protocol state machines and one agent-security position paper — with no single field dominant, so reading it as unfocused would be premature at this catalog depth.
Collaboration66/100
Collaboration scores 66, backed by 27 distinct coauthors in the snapshot and a 12-name frequent-coauthor list spanning both of her fields, including Anqi Chen, Dan Goldwasser, Eugene Bagdasarian, Luca Beurer-Kellner, Battista Biggio, and Heather Frase. Both catalog papers are team efforts — middle authorship on the position paper and last authorship on RFCLLM — consistent with senior placement, though the corresponding-author flag is absent for both.
Activity—
Not scored: the snapshot provides no per-year publication cadence, so recent output rate cannot be computed. This axis would measure publishing consistency and momentum over time — papers per recent year, gaps, and acceleration — beyond the two 2026 entries available here.
Reproducibility60/100
Reproducibility scores 60, the mean across 2 papers carrying readiness assessments. The split is instructive: RFCLLM rates high — a benchmark grounded in RFC-specified protocol state machines that others can rerun — while the agent incident-reporting position paper rates low, a conceptual framework with neither a solicitation method nor validation. The 60 average reflects one empirically solid artifact alongside one position piece rather than uniform practice.
Research summary
Cristina Nita-Rotaru (Northeastern University) is profiled here from public bibliographic data and Evolve's reviews of two 2026 arXiv preprints, both sitting at the intersection of AI security and LLM-based analysis of networked systems. Her bibliometric field split is exactly even — 0.5 for natural language processing (LLM benchmarking for network protocol analysis) and 0.5 for AI security and governance — which yields a focus score of 0 and signals deliberate dual-track positioning rather than concentration. RFCLLM, where she is last author, asks whether LLMs can reason over finite state machines distilled from RFC specifications; Evolve judges the benchmark credible and its reproducibility readiness high, while flagging single-annotator gold answers and judge-model overlap. Beyond Predictable Paths, where she is a middle author, argues that security incident reporting must be redefined for autonomous agents around traces, autonomy, delegation and identity, and multi-agent data; Evolve finds the framework clear but lacking a solicitation method or validation. The snapshot records 27 distinct coauthors (collaboration score 66) and a reproducibility mean of 60 across 2 papers, with frequent coauthors spanning both fields. Output, impact, and activity axes remain unscored for insufficient data, so treat this as an early-stage, catalog-scoped profile.
Collaboration pattern
The snapshot records 27 distinct coauthors and a collaboration score of 66. The 12-name frequent-coauthor list spans both active fields — AI security and governance (e.g., Battista Biggio, Eugene Bagdasarian, Heather Frase) and LLM/NLP benchmarking (e.g., Dan Goldwasser, Luca Beurer-Kellner) — plus recurring names such as Anqi Chen, Anastasia Pustozerova, Nico Ebert, and Marc Fischer. On the two catalog papers she holds middle and last author positions, neither as corresponding author, suggesting team-based rather than solo work.
Research keywords
Highlights
RFCLLM: LLMs struggle to reason over protocol state machines
High reproducibility readiness for the protocol benchmark
Redefining security incident reporting for AI agents
Timeline
- 2026Posts RFCLLM on arXiv, a benchmark evaluating LLMs' reasoning over network protocol state machines distilled from RFC specifications (last-author position).
- 2026Posts Beyond Predictable Paths on arXiv, a position paper redefining AI security incident reporting for agents (middle-author position).
- 2026Both papers enter the Evolve catalog, establishing an even 50/50 split between LLM protocol benchmarking and AI security governance.
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.
- Radar scores are computed by Evolve from public bibliometric aggregation; output, impact, and activity lack sufficient underlying data and are intentionally unscored.
- Profile scope is limited to two 2026 arXiv preprints reviewed by Evolve plus their co-author metadata; earlier publications, venue acceptance, and career history are not captured.
- Both catalog papers are preprints; peer-review status and final publication venues are not reflected in this snapshot.
- Name resolution relies on the provided ORCID (0000-0002-9649-6789) and Northeastern University affiliation; other researchers with similar names are out of scope.