Ohio State data-mining researcher expanding into federated-learning incentives and speech-compression fairness
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
67
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-15) 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 is not yet scored. This axis would aggregate a validated, de-duplicated work corpus: total works, annual publication counts and authorship positions. The snapshot lacks that corpus for this identity, so no score or estimate is offered.
Impact—
Impact is not yet scored. This axis would use citation totals, yearly citation trajectories and venue influence; the snapshot has no verified citation corpus, and identity noise undermines the reported figures, so no score is offered.
Focus0/100
The focus score is 0, the widest spread on the radar. The OpenAlex topic profile spans five similarly sized clusters — Data Mining Algorithms and Applications (99), Advanced Database Systems and Queries (89), Complex Network Analysis Techniques (78), Graph Theory and Algorithms (69) and Enzyme Structure and Function (68) — and the two reviewed 2026 papers sit in unrelated fields (speech-recognition compression; federated-learning mechanism design), confirming breadth over niche concentration.
Collaboration36/100
Collaboration scores 36, a moderate reading: the recent window lists only 5 distinct coauthors despite 117 works across three years. Recurring partners include Amin Meghrazi, Andrew Perrault, Srishti Ginjala, Eric Fosler-Lussier and Christopher W. Myers; on both reviewed 2026 papers the author position is middle or last and non-corresponding, consistent with a senior collaborator joining graduate-led projects.
Activity100/100
Activity is the standout axis at 100. The record logs 117 works in the most recent three-year window, and countsByYear climbs from 29 works in 2023 to 40 in 2024 and 54 in 2025, with 23 already recorded for 2026 — an accelerating, currently peak output rate within the provided data.
Reproducibility55/100
Reproducibility scores 55 (reproMean 55 across 2 reviewed papers). Both Evolve-reviewed 2026 preprints carry a medium repro-readiness rating: the federated-learning paper is noted for clean theory with honest three-seed empirics, while the Whisper compression audit is rigorous but confined to data-free 4-bit recipes and read speech — solid, but not turnkey replication.
Research summary
Srinivasan Parthasarathy is a US-based researcher at The Ohio State University, name-resolved here from OpenAlex public bibliographic data (ORCID 0000-0002-6062-6449). The record shows a long, high-volume career: an h-index of 67 and i10-index of 242, with 117 works in the most recent three-year window and countsByYear rising from 29 works (2023) to 40 (2024) and 54 (2025) — the basis for a maximal activity score of 100. The topic profile is unusually broad — Data Mining Algorithms and Applications (99), Advanced Database Systems and Queries (89), Complex Network Analysis Techniques (78), Graph Theory and Algorithms (69) and Enzyme Structure and Function (68) — which yields a focus score of 0. Evolve has reviewed two 2026 arXiv papers: a federated-learning mechanism-design study formalizing pathwise individual rationality (middle authorship), and an audit of Whisper post-training compression showing error shifts toward marginalized speakers (last authorship). Both carry medium repro-readiness, grounding a reproducibility score of 55 (reproMean over 2 papers). Collaboration scores a moderate 36, with 5 distinct recent coauthors and recurring partners including Amin Meghrazi, Andrew Perrault, Srishti Ginjala, Eric Fosler-Lussier and Christopher W. Myers. Output and impact dimensions are marked missing; the record's pre-1950 publication years also flag entity-resolution noise, so aggregate figures should be read with care.
Collaboration pattern
The recent window shows 5 distinct coauthors against 117 works, and collaboration scores a moderate 36. Recurring partners are Amin Meghrazi, Andrew Perrault, Srishti Ginjala, Eric Fosler-Lussier and Christopher W. Myers. On both Evolve-reviewed 2026 papers Parthasarathy takes a middle or last, non-corresponding position, suggesting a senior-collaborator pattern: contributing methods and supervision to graduate- or junior-led projects rather than leading small self-led teams. The low distinct-coauthor count may partly reflect the recent-window snapshot.
Research keywords
Highlights
Pathwise individual rationality for federated learning
Compression shifts Whisper's errors onto marginalized speakers
Recent output surge
countsByYear climbs from 29 works in 2023 to 40 in 2024 and 54 in 2025, with 117 works across the recent three-year window and 23 already logged for 2026 — the basis for the maximal activity score of 100.
Timeline
- 2006OpenAlex countsByYear logs 43 works — the highest single-year figure in the provided record, within a strong 2005–2008 run (26, 43, 22, 25).
- 2023countsByYear records 29 works, starting an acceleration to 40 in 2024 and 54 in 2025 — the strongest recent stretch, underpinning the activity score of 100.
- 2026Evolve reviews "Pathwise Individual Rationality in Federated Learning: A Mechanism-Architecture Co-Design" (arXiv), with this researcher in a middle, non-corresponding author position.
- 2026Evolve reviews "Temporal Taxation Compounds Under Post-Training Compression of Whisper Models" (arXiv), with this researcher as last author; the study finds compression shifts error burden onto marginalized speakers.
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.
- All bibliometrics come from the OpenAlex record; the countsByYear series includes works dated 1936–1990, which strongly suggests entity-resolution noise, so h-index 67, i10-index 242 and yearly counts may be inflated.
- Citation-based output and impact scores are missing (dimMissing): no validated citation corpus was available for radar scoring.
- Evolve review coverage is limited to two 2026 arXiv preprints; the fieldDist split (0.5 / 0.5) is derived from those papers and may not represent the full career output.
- Identity is resolved from public bibliographic data (OpenAlex, ORCID 0000-0002-6062-6449); confidence depends on that resolution.