Meghanadh Pulivarthi

Indian Institute of Technology Madras · India

Unclaimed · AI-structured profileOpenAlex

NLP researcher at IIT Madras working on knowledge injection and structure-aware retrieval for large language models

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-17) 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: the snapshot lacks a computable publication-volume record — only 4 works appear, all from 2026, with no fuller catalog of prior outputs. This axis would measure cumulative and per-year publication counts across venues and preprint servers.

Impact—

Not scored: h-index and i10-index are both 0 and no citation counts are available yet. This axis would measure citation-based influence — h-index, i10-index and per-paper citations — which is premature to compute for this 2026-heavy record.

Focus0/100

The focus axis reads 0 on this snapshot, even though the underlying field distribution is narrow: it splits exactly 50/50 between NLP (LLM adaptation and knowledge injection) and NLP (information retrieval). Topic counts concentrate in Topic Modeling (4), Information Retrieval and Search Behavior (3), and Multimodal Machine Learning Applications (3), with NLP Techniques and Domain Adaptation and Few-Shot Learning at 1 each.

Collaboration50/100

Scores 50. OpenAlex records 11 distinct coauthors over the recent three-year window, anchored by a recurring cluster: Vineet Kumar, Jaydeep Sen, Sachindra Joshi, Kushagra Bhushan, Gaurav Pandey, Yatin Nandwani, Dinesh Raghu, Sai Krishna Reddy Sathi, Sonam Gupta, Vishwajeet Kumar and Riyaz Ahmad Bhat. In the catalog he is a middle author on STAIR and DKL and first author on KItCAT, pointing to steady team-based NLP research.

Activity47/100

Scores 47, with a short visible window: countsByYear contains only 2026, with 4 works, matching recent3yWorks of 4. All three catalog papers are 2026 arXiv preprints (STAIR, DKL, KItCAT), so recorded activity is compressed into a single year; longer-term productivity trends cannot be assessed from this snapshot.

Reproducibility55/100

Scores 55, from a reproMean of 55 across 2 repro-scored papers. The two reviewed papers, STAIR and DKL, both carry medium reproducibility readiness: each reports concrete benchmark numbers — STAIR at 82.6 vs 76.9 R@1 over DSI — while Evolve's verdicts flag caveats such as synthetic query generation, constrained decoding, and comparison margins within noise.

Research summary

Meghanadh Pulivarthi is a natural language processing researcher affiliated with the Indian Institute of Technology Madras, India. This profile is name-resolved from OpenAlex bibliographic data; no ORCID is on file. The visible record is recent and compact: three 2026 arXiv preprints, all in the Evolve catalog. As first author he led KItCAT (Knowledge Injection via Input Corruption for Auto-regressive Training). As a middle author he contributed to STAIR, which makes a book's table of contents the retrieval unit of a generative retriever and is reported to beat DSI (82.6 vs 76.9 R@1), and to DKL, which merges a base-LLM knowledge vector into an instruction-tuned model with strong data-efficiency and QA gains, though Evolve notes the leads over RAG baselines are often within noise. OpenAlex field weights split evenly (0.5/0.5) between LLM adaptation and knowledge injection and information retrieval, and topic counts concentrate in topic modeling, IR and search behavior, and multimodal machine learning. He has 11 distinct coauthors in the recent three-year window, with a tight recurring collaborator cluster. Citation indicators (h-index and i10-index, both 0) leave the output and impact axes unscored, so this is an early-stage profile whose substance rests on the reviewed 2026 preprints rather than on citation trace.

Collaboration pattern

OpenAlex counts 11 distinct coauthors in the recent three-year window, and the frequent-coauthor list forms a tight recurring cluster: Vineet Kumar, Jaydeep Sen, Sachindra Joshi, Kushagra Bhushan, Gaurav Pandey, Yatin Nandwani, Dinesh Raghu, Sai Krishna Reddy Sathi, Sonam Gupta, Vishwajeet Kumar and Riyaz Ahmad Bhat. Authorship in the catalog is team-based — middle author on STAIR and DKL, first author on KItCAT — consistent with collaborative lab-style NLP research.

Research keywords

retrieval-augmented generationdifferentiable search indexknowledge injectionmodel mergingtask arithmeticinstruction tuningextended pre-trainingparametric memorydocument structuregenerative retrieval

Highlights

STAIR: table of contents as the retrieval unit
Evolve's review: STAIR makes a book's ToC the retrieval unit of a generative retriever, beating DSI at 82.6 vs 76.9 R@1; the verdict notes that synthetic queries and constrained decoding temper the claims. Reproducibility readiness: medium.
DKL: decoupled knowledge learning for instruction-tuned LLMs
DKL merges a base-LLM knowledge vector into an instruct LLM; Evolve finds the data-efficiency and QA gains strong, but the leads over the best RAG baselines are often within noise. Reproducibility readiness: medium.
KItCAT: first-authored knowledge injection work
Knowledge Injection via Input Corruption for Auto-regressive Training is Pulivarthi's first-author 2026 arXiv preprint; Evolve has not yet published a review, keywords, or reproducibility assessment for this paper.

Timeline

2026Co-authored the STAIR preprint on arXiv, a structure-aware information retriever that treats a book's table of contents as the retrieval unit.
2026Co-authored the DKL preprint on arXiv, decoupled knowledge learning that merges base-LLM knowledge into instruction-tuned models.
2026First-authored the KItCAT preprint on arXiv, knowledge injection via input corruption for auto-regressive training.

Papers in the Evolve catalog

Frequent co-authors

Vineet Kumar×3 Jaydeep Sen×3 Sachindra Joshi×3 Kushagra Bhushan×2 Gaurav Pandey×2 Yatin Nandwani×2 Dinesh Raghu×2 Sai Krishna Reddy Sathi×1 Sonam Gupta×1 vishwajeet kumar×1 Riyaz Ahmad Bhat×1

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 indicators come from OpenAlex only; h-index and i10-index are both 0, so output and impact are reported as missing rather than scored.
The visible record covers three 2026 arXiv preprints; no peer-reviewed venues, funding, or award data are in scope.
KItCAT has no Evolve review, keywords, or reproducibility assessment yet, and no catalog slug.
ORCID is not on file; the identity is name-resolved from public OpenAlex bibliographic data.