NLP and agent-safety researcher; first author of 2026 arXiv work on scam-call screening and agentic PDF forgery
2
papers here
2
reviewed by Evolve
2
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-09) 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: this profile has no publication-volume or productivity data. The axis would measure how many papers Simiao Ren produces over time, in which venues, and in what author positions.
Impact—
Not yet scored: citation counts, h-index, and downstream uptake are absent from the current snapshot. The axis would measure how often and how widely this work is cited and built upon by others.
Focus0/100
The focus axis scores 0, and the raw field distribution explains it: catalogued output splits exactly evenly (0.5 and 0.5) between natural language processing and AI agent safety & document forensics. With just two 2026 first-author papers, one per field, this reads as a deliberate two-track agenda — efficient small-model screening on one side, agentic document-forgery evaluation on the other — rather than a settled single specialization.
Collaboration48/100
Collaboration scores 48, grounded in 10 distinct coauthors across the catalogued work. Simiao Ren takes first-author, non-corresponding positions on both 2026 papers, and the frequent-coauthor list — Ankit Raj, Tommy Duong, Yuxin Zhang, Dennis Ng, Xingyu Shen, Kidus Zewde, Yuchen Zhou, Neo Tiangratanakul — points to a recurring working group of moderate size. The mid-range score suggests steady teamwork without yet a large or highly distributed collaboration network.
Activity—
Not yet scored: temporal signals such as publication cadence and recent-year output are missing from the snapshot. The axis would measure how consistently Simiao Ren publishes and stays research-active over time.
Reproducibility55/100
Reproducibility scores 55 (reproMean 55), computed from 2 reviewed papers, and Evolve rates both as medium reproducibility readiness. The scam-screening paper is candid about test-set exposure but relies entirely on synthetic data; the PDF-forgery benchmark uses a rigorous rule-based scoring design, yet server-side heterogeneity confounds model rankings. The mid-range score reflects careful evaluation methodology paired with data or environment limitations in each paper.
Research summary
Simiao Ren is name-resolved from public bibliographic data; no institution, country, or ORCID is attached to this profile. The catalogued record centers on two first-author arXiv preprints, both dated 2026 and split evenly (0.5/0.5) between natural language processing and AI agent safety & document forensics. Open-Jev Judgments on CallScreenBench shows that a JevLite typed-decision readout fine-tuned on Qwen3-4B performs calibrated one-pass scam-call screening non-inferior to an LLM judge at roughly 30x lower latency; Evolve credits its candor about test-set exposure but notes all data is synthetic. Agents That Edit Documents measures how reliably LLM agents forge archived PDFs against a non-agentic control, reporting an 81.1% raw success rate that falls to 46.2% under strict filtering; Evolve finds the rule-based scoring rigorous while flagging server-side heterogeneity as a ranking confound. Radar scores place collaboration at 48 (10 distinct coauthors), reproducibility at 55 (reproMean across 2 papers, both medium readiness), and focus at 0, mirroring the even two-field split; output, impact, and activity metrics are not yet available. Read this as an early-stage profile: recent, first-authored, preprint-phase work with unusually careful attention to calibration and evaluation methodology.
Collaboration pattern
Simiao Ren appears as first author (not corresponding) on both catalogued 2026 papers, working with a pool of 10 distinct coauthors; the collaboration radar score is 48. Recurring names include Ankit Raj, Tommy Duong, Yuxin Zhang, Dennis Ng, Xingyu Shen, Kidus Zewde, Yuchen Zhou, and Neo Tiangratanakul, suggesting a stable but moderately sized team rather than a large distributed network. Both collaboration lines stay within the same two-field scope: NLP screening and agentic document forensics.
Research keywords
Highlights
Small-model scam screening matches an LLM judge at ~30x lower latency
Quantifying how reliably LLM agents forge archived PDFs
Timeline
- 2026First catalogued output: posts 'Open-Jev Judgments on CallScreenBench' to arXiv as first author, introducing calibrated one-pass scam-call screening with a small language model (Qwen3-4B).
- 2026Publishes 'Agents That Edit Documents' on arXiv as first author, benchmarking agentic PDF forgery against a non-agentic control (81.1% raw, 46.2% strict-filtered success rates).
- 2026Evolve reviews both preprints in the same year, rating each medium on reproducibility readiness and computing a reproducibility score of 55 across the 2 papers.
- 2026The bibliometric snapshot records an even 0.5/0.5 split between NLP and AI agent safety & document forensics, with 10 distinct coauthors across the catalogued work.
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.
- Bibliometrics and radar scores are computed from the provided bibliometric snapshot only; the output, impact, and activity dimensions lack underlying data.
- No institution, country, or ORCID is attached; the profile is name-resolved from public bibliographic data, and 'Simiao Ren' may not be a unique name.
- Both catalogued papers are 2026 arXiv preprints; peer-reviewed venue, citation, and funding information are out of scope.
- The reproducibility score (55) averages only 2 reviewed papers, both carrying synthetic-data or evaluation-heterogeneity limitations flagged by Evolve.