NLP researcher building governed document AI — versioning, deduplication, and LLM artifact generation for enterprises
2
papers here
2
reviewed by Evolve
2
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-09-21) 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: Evolve holds no aggregate publication-volume data for this author beyond the two reviewed 2026 arXiv preprints. This axis would measure total scholarly output across venues and years, including preprints, conference papers, and journal articles.
Impact—
Not yet scored: no citation counts, h-index, or download statistics appear in the available data. This axis would measure the influence of the two 2026 preprints — citations accrued over time and uptake within the document-AI community.
Focus0/100
Scored 0/100, the lowest reading, while the observed field distribution splits evenly (0.5 each) between natural language processing and document AI/applied LLM systems. The two directions are thematically adjacent — both governed enterprise document pipelines — so genuine concentration is plausible, but with only two cataloged papers the per-field evidence is too thin to support a meaningful focus score.
Collaboration44/100
Scored 44/100 from 8 distinct coauthors across the two catalog papers. Sediqin takes first authorship on both, working with a compact recurring core — Shivali Dalmia, Sumukha Thoppanahalli, Abhishek Mukherji, and Srinivasa Karthikeya Reddy Kovvuri — suggesting a small, stable team rather than a broad distributed network. With only two papers observed, the moderate score reflects a limited observable sample more than a weak collaboration record.
Activity—
Not yet scored: the input data records only the year 2026 for both papers, with no publication history before or after. This axis would measure publishing cadence and sustained research activity over time, which cannot be assessed from a single-year snapshot.
Reproducibility20/100
Scored 20/100, the mean repro-readiness across the 2 reviewed papers, both rated low. GVD's strong F1 comes from a private enterprise corpus and is described as not independently reproducible, while SAGE rests on proprietary data with proxy metrics such as LLM-as-judge. Releasing code, corpora, or synthetic evaluation sets would materially raise this axis.
Research summary
Mohammadreza Sediqin is a natural language processing researcher working at the intersection of document AI and applied LLM systems, with an emphasis on governed, enterprise-grade document pipelines. The Evolve catalog reviews two of his 2026 arXiv preprints, both with him as first author. GVD introduces governed versioning and deduplication for document repositories, unifying version-family linking with rule-level conflict governance — notably without relying on LLMs — and reports strong F1 on a private enterprise corpus. SAGE targets guideline-to-artifact generation through a governed LLM pipeline with human-in-the-loop oversight, cutting the work from days to roughly one hour at a reported ~3.2% hallucination rate, validated via LLM-as-judge and proxy metrics. His observed field distribution splits evenly (0.5/0.5) between core NLP and document AI/applied LLM systems, and he works with a compact pool of 8 distinct coauthors, including Shivali Dalmia, Sumukha Thoppanahalli, Abhishek Mukherji, and Srinivasa Karthikeya Reddy Kovvuri. The main limitation is evaluation transparency: both papers are rated low on reproducibility readiness (mean 20) because they rest on proprietary data. Aggregate output, citation impact, and multi-year activity data are not yet available, making this an early-stage profile.
Collaboration pattern
Sediqin publishes as first author within a compact recurring team: Shivali Dalmia, Sumukha Thoppanahalli, Abhishek Mukherji, and Srinivasa Karthikeya Reddy Kovvuri appear alongside him on the cataloged work. Across the two 2026 papers he draws on 8 distinct coauthors in total — consistent with a small, stable, enterprise-style research group rather than a broad academic network. No corresponding-author role is recorded for him in the reviewed papers.
Research keywords
Highlights
Governed versioning and deduplication — without LLMs
SAGE compresses guideline-to-artifact work from days to about one hour
A consistent reproducibility gap
Both 2026 preprints receive low repro-readiness ratings because evaluations depend on proprietary enterprise corpora — a shared limitation that anchors the profile's overall reproducibility score of 20 and is the clearest open issue in this body of work.
Timeline
- 2026First-author arXiv preprint GVD proposes governed versioning and deduplication for document repositories, unifying version-family linking and rule-level conflict governance without LLMs.
- 2026First-author arXiv preprint SAGE introduces governed artifact generation from enterprise guidelines via a human-in-the-loop LLM pipeline, reporting ~1-hour turnaround at ~3.2% hallucination.
- 2026Both papers enter the Evolve catalog with low repro-readiness ratings, anchoring an early-stage profile in governed document AI and applied LLM systems.
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 derive from a limited public bibliographic snapshot; only two Evolve-reviewed papers underpin this profile.
- No institution, country, or ORCID was available; the profile is name-resolved from public bibliographic data.
- Output, impact, and activity axes are unscored because aggregate publication, citation, and multi-year volume data are missing.
- Both papers are 2026 arXiv preprints evaluated on proprietary enterprise corpora, limiting independent verification.