GPU retrieval and pre-ranking engineer at LinkedIn; recommender-systems specialist with broad industrial collaboration
2
papers here
1
reviewed by Evolve
0
first-author
0
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 yet: the supplied snapshot lists only two catalog papers and no publication counts. This axis would measure cumulative output volume and its year-by-year distribution, which cannot be estimated from current data.
Impact—
Not scored yet: no citation counts, h-index, or downstream adoption metrics were provided. This axis would measure how widely the work is cited and deployed, so it is left unscored rather than estimated.
Focus100/100
Scored 100: the field distribution in the available record maps entirely to a single field — recommender systems / information retrieval (1 of 1 field-tagged paper). Both catalog papers, on GPU pre-ranking at LinkedIn and efficient GPU retrieval for semantic search, sit in the same technical neighborhood of retrieval, ranking, and GPU inference, indicating tightly concentrated specialization.
Collaboration66/100
Scored 66 from a raw distinct-coauthor count of 27 in the available record. Twelve recurring collaborators — including Dhritiman Das, Chujie Zheng, Ronak Kaoshik, Pratik Dixit, Vishal Shah, and Yanbo Li — suggest a stable industrial team at LinkedIn rather than one-off ties. The reviewed papers list the researcher in a middle author position, consistent with contributor roles inside a large engineering group rather than solo leadership.
Activity—
Not scored yet: the snapshot gives no per-year publication cadence beyond two 2026 papers. This axis would measure recent publishing intensity and continuity, and it stays unscored until a longer publication history is verifiable.
Reproducibility20/100
Scored 20, computed from a reproducibility mean of 20 across 1 assessed paper. The one reviewed paper with a readiness rating — the Connected Content Retriever — is rated low: Evolve's verdict notes the system lifts Feed time spent by 2.5% but that training and baseline details are thin. Industrial systems work of this kind is often constrained by proprietary data and infrastructure.
Research summary
Chirag Bhanuprasad Mehta is a researcher at LinkedIn (United States) whose visible record centers on GPU-accelerated retrieval and ranking for large-scale recommender systems. The Evolve catalog holds two 2026 arXiv papers, both with middle authorship and no corresponding-author role: 'Connected Content Retriever: Dense Graph Edge Features Powering Pre-Ranking at LinkedIn' and 'Efficient GPU Retrieval for Semantic Search'. The first is fully reviewed: Evolve describes a GPU pre-ranker that joins dense graph edge features at runtime, lifts Feed time spent by 2.5%, and calls it solid systems work while noting that training and baseline details are thin; reproducibility readiness is rated low. The second extends the GPU-retrieval theme to semantic search but has no verdict, keywords, or readiness rating yet. The field distribution in the record maps entirely to recommender systems / information retrieval, yielding a focus score of 100. Collaboration spans 27 distinct coauthors, with a recurring core of twelve, including Dhritiman Das, Chujie Zheng, Ronak Kaoshik, Pratik Dixit, Vishal Shah, and Yanbo Li. Reproducibility scores 20, based on one assessed paper. Output, impact, and activity are unscored because no publication counts, citation metrics, or multi-year cadence were provided, so this should be read as an early-stage, industry-focused profile resolved from public bibliographic data.
Collaboration pattern
The available record shows 27 distinct coauthors, with a dense recurring core of twelve, including Dhritiman Das, Chujie Zheng, Ronak Kaoshik, Pratik Dixit, Vishal Shah, Yanbo Li, Jiahao Xu, Manika Agarwal, Chinmay Naik, Lingyu Zhang, Chetan Bhole, and Meng Zheng. Both catalog papers carry middle authorship, suggesting this person operates as a contributing engineer within large LinkedIn retrieval teams rather than as a corresponding or sole lead.
Research keywords
Highlights
Runtime edge-feature joins in GPU pre-ranking lift Feed time spent +2.5%
A second 2026 paper extends GPU retrieval to semantic search
'Efficient GPU Retrieval for Semantic Search' carries middle authorship like the LinkedIn pre-ranking paper, but has no Evolve verdict, keywords, or reproducibility rating yet, so its contribution remains thinly documented.
Reproducibility readiness rated low for the flagship paper
Timeline
- 2026Published 'Connected Content Retriever: Dense Graph Edge Features Powering Pre-Ranking at LinkedIn' on arXiv, describing a GPU pre-ranker with runtime edge-feature joins.
- 2026Published 'Efficient GPU Retrieval for Semantic Search' on arXiv, extending the GPU-retrieval line toward semantic search.
- 2026Evolve reviewed the LinkedIn pre-ranking paper: Feed time spent +2.5%, judged solid systems work, with reproducibility readiness rated low.
- 2026Collaboration footprint in the record reaches 27 distinct coauthors, including twelve frequent collaborators at LinkedIn.
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 scores are derived from an OpenAlex-based snapshot; only two catalog papers were available for this researcher.
- No ORCID or country field was supplied; identity is name-resolved from public bibliographic data and this name may be shared by multiple people.
- Output, impact, and activity are unscored because no publication counts, citation metrics, or multi-year cadence were provided.
- One of the two papers has no Evolve verdict, keywords, field tag, or reproducibility rating yet.