NLP researcher building governed enterprise document AI — LLM-free versioning/dedup and human-governed LLM pipelines
4
papers here
4
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
6
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) 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—
Output volume — paper counts by year and venue — cannot yet be scored: the record covers only the two Evolve-reviewed 2026 arXiv preprints, with no full publication list or venue history available.
Impact—
Impact — citations, h-index, downstream adoption — is unscored because no citation data was provided; both catalogued preprints are recent 2026 arXiv postings whose evaluations rest on proprietary corpora, so external uptake is unmeasured.
Focus0/100
Scored 0/100, reflecting a field distribution split evenly (0.5/0.5) between general natural language processing and NLP (document AI / applied LLM systems). The two labels are adjacent rather than divergent — both 2026 papers target governed enterprise document processing — so the low score captures formal label dispersion more than genuine topic scatter.
Collaboration54/100
Scored 32/100 from 4 distinct coauthors across the catalogued work. Both 2026 papers are written with the same compact group — Mohammadreza Sediqin, Shivali Dalmia, Sumukha Thoppanahalli, and Srinivasa Karthikeya Reddy Kovvuri — with Mukherji in the last author position each time, a pattern pointing to a stable small-team collaboration style and a senior role rather than broad external networking.
Activity91/100
Activity — publication cadence over time — is not yet scoreable: the record holds only two 2026 preprints, so annual output trends and sustained research tempo cannot be established.
Reproducibility38/100
Scored 20/100, the mean repro-readiness across 2 reviewed papers, both rated low. GVD reports strong F1 on a private enterprise corpus and SAGE's evaluation rests on proprietary data and proxy metrics, so neither currently supports independent replication — releasing corpora, rules, and pipeline configurations would be the clearest lever to raise this score.
Research summary
Abhishek Mukherji is a researcher working at the intersection of natural language processing and enterprise document AI; the catalog record to date comprises two 2026 arXiv preprints reviewed by Evolve: GVD (Governed Versioning and Deduplication for Document Repositories) and SAGE (Governed Artifact Generation from Enterprise Guidelines). On both he takes the last author position — typically a senior or supervisory placement — working with a compact recurring team of four co-authors: Mohammadreza Sediqin, Shivali Dalmia, Sumukha Thoppanahalli, and Srinivasa Karthikeya Reddy Kovvuri. Both papers share one organizing theme: governance. GVD unifies version-family linking and rule-level conflict governance without relying on LLMs, reporting strong F1 on a private enterprise corpus. SAGE applies a governed LLM pipeline with human-in-the-loop checks to generate artifacts from enterprise guidelines, cutting turnaround from days to roughly one hour at a reported 3.2% hallucination rate. Evolve rates both papers' reproducibility readiness as low, since evaluation rests on proprietary data and proxy metrics. The field distribution across the catalog splits evenly between general NLP and NLP (document AI / applied LLM systems). Because no citation record, full publication list, or affiliation data is available, output, impact, and activity dimensions remain unscored, and this profile should be read as an early-stage, name-resolved view built from public bibliographic data.
Collaboration pattern
Mukherji works with a tight, recurring four-person core — Mohammadreza Sediqin, Shivali Dalmia, Sumukha Thoppanahalli, and Srinivasa Karthikeya Reddy Kovvuri — accounting for all 4 distinct coauthors in the catalog. He takes the last author position on both 2026 papers without being marked corresponding, suggesting a senior but hands-off supervisory role within a stable small team rather than a wide external network.
Research keywords
Highlights
Governed versioning for enterprise document repositories
From days to ~1 hour: governed artifact generation
A consistent governance-first research program
Across both 2026 papers — one LLM-free, one LLM-powered — the same question recurs: how to put auditable rules, version control, and human oversight around enterprise document workflows, spanning near-duplicate detection, NLI-based contradiction checks, and LLM-as-judge evaluation.
Timeline
- 2026GVD preprint posted on arXiv, introducing governed versioning and deduplication for document repositories with Mukherji as last author.
- 2026SAGE preprint posted on arXiv, applying a governed LLM pipeline to generate artifacts from enterprise guidelines, again with Mukherji as last author.
- 2026Both catalogued papers share the same four-person coauthor core, establishing a document-AI governance focus within a stable small team.
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 come from a thin public bibliographic snapshot: two 2026 arXiv preprints and 4 distinct coauthors; no venue history or citation record is included.
- No ORCID, institution, or country is attached, so this profile is name-resolved from public bibliographic data and identity is not independently confirmed.
- Output, impact, and activity dimensions are unscored because citation counts and full publication lists were absent from the input data.
- Per Evolve's review verdicts, both papers evaluate on proprietary enterprise corpora and proxy metrics, limiting independent verification of reported results.