Industrial process ML veteran at TCS pivoting to graph-grounded multimodal and agentic AI systems
4
papers here
1
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
20
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—
No output score is available yet: verified aggregate publication and productivity metrics are missing from this snapshot. This axis would measure total works, the full output trajectory, and venue spread across the career record once a complete bibliometric feed is loaded for this researcher.
Impact—
No impact score is computed yet. The raw record does list h-index 20 and i10-index 31, but citation aggregates are too incomplete to score influence reliably. This axis would measure citation uptake, citing-work quality, and scholarly attention across the researcher's body of work.
Focus100/100
Focus scores 100 on a highly concentrated topic profile. OpenAlex-derived clusters put fault detection and control systems first with 27 works, followed by iron and steelmaking processes (15), model reduction and neural networks (14), multimodal machine-learning applications (13), and machine learning in materials science (12) — a coherent process-industries ML agenda now extending into diagram understanding.
Collaboration22/100
Collaboration scores 22, the lowest axis. The recent window records only 2 distinct coauthors against 69 works in three years, an unusually small direct network for that volume. The 2026 arXiv series repeatedly pairs Runkana (last-listed author) with Sagar Srinivas Sakhinana and Prathamesh Gadekar, indicating a tight in-house group rather than a broad external collaboration web.
Activity100/100
Activity scores 100. The year-by-year record shows output since 2000 that accelerates sharply in recent years: 7 works in 2021, 13 in 2022, 13 in 2023, a peak of 38 in 2024, then 15 in 2025 and 16 in 2026. The last three years account for 69 works, indicating a currently very high publication tempo.
Reproducibility55/100
Reproducibility scores 55 (reproMean), drawn from a single repro-reviewed paper: the grounded P&ID reasoning study, rated medium reproducibility readiness. Its reported gains of 74.3–76.0% over 36.7–41.3% baselines are measured on synthetic sheets with harness factors unablated. With only one assessed paper, this score should be read as preliminary evidence.
Research summary
Venkataramana Runkana is a researcher at Tata Consultancy Services (India) whose record spans two decades of industrial process engineering and a recent, intensive move into machine-learning systems research. The OpenAlex-derived topic profile clusters into five areas: fault detection and control systems (27 works), iron and steelmaking processes (15), model reduction and neural networks (14), multimodal machine-learning applications (13), and machine learning in materials science (12). Activity metrics show a steep ramp: 7 works in 2021 rising to a peak of 38 in 2024, with 69 works in the last three years and 15 (2025) and 16 (2026) recorded so far. The Evolve catalog lists four 2026 arXiv papers, all with Runkana in the last author position: a grounded P&ID reasoning study constraining vision-language models with recovered evidence graphs, two companion papers on zero-trust agentic data engineering and cloud engineering, and a paper on forward-deployed full-stack engineering for autonomous cloud MLOps. The reviewed P&ID paper reports graph grounding lifting P&ID topology QA from 36.7–41.3% to 74.3–76.0% on synthetic sheets, with medium reproducibility readiness; Evolve finds the lift convincing but synthetic-only, with harness factors unablated. The raw record lists h-index 20 and i10-index 31. Distinct coauthors in the recent window number just two, with Sagar Srinivas Sakhinana and Prathamesh Gadekar recurring across the 2026 series.
Collaboration pattern
Runkana sits last-listed (not corresponding) on all four cataloged 2026 papers, a senior-supervisor signature. The recent window records just 2 distinct coauthors across 69 works; the recurring partners are Sagar Srinivas Sakhinana and Prathamesh Gadekar, who co-sign the P&ID and agentic-engineering series. The pattern suggests a small, stable industrial research group rather than wide academic networking, consistent with the collaboration score of 22.
Research keywords
Highlights
Graph grounding lifts P&ID topology QA from 36.7–41.3% to 74.3–76.0%
A 2026 arXiv series on zero-trust agentic engineering
Three companion papers — zero-trust agentic data engineering and analytical processing, agentic cloud engineering with a graph-and-loop agent harness, and forward-deployed full-stack engineering for autonomous cloud MLOps — carry the same graph, loop, and harness engineering recipe to cloud-scale automation; Evolve review metadata for these is not yet available.
Two decades from process-industry ML to agentic AI
The OpenAlex topic clusters trace a line from fault detection and control systems (27 works) and iron and steelmaking processes (15) through model reduction and neural networks (14) to multimodal machine-learning applications (13) and machine learning in materials science (12), with the 2026 arXiv series marking a pivot toward agentic systems.
Timeline
- 2000Earliest recorded publication year in the OpenAlex year-by-year record, with 1 work.
- 2003Early-career output of 3 recorded works, rising to 4 in 2005.
- 2021Output ramps up to 7 recorded works, beginning a steep multi-year climb.
- 2024Peak recorded year with 38 works in the OpenAlex counts.
- 202515 works recorded as the research agenda shifts toward agentic and multimodal systems.
- 202616 works recorded so far; four arXiv papers enter the Evolve catalog, all last-authored, including the grounded P&ID reasoning study and the zero-trust agentic engineering series.
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, topic clusters, and year-by-year counts are derived from an OpenAlex snapshot; citation coverage and author disambiguation may be incomplete.
- Output and impact radar axes are unscored in this profile because verified aggregate data is insufficient; h-index 20 and i10-index 31 are raw record values, not platform scores.
- Three of the four cataloged 2026 arXiv papers have no Evolve verdict, field, keywords, or reproducibility assessment yet.
- Profile is name-resolved from public bibliographic data (OpenAlex; ORCID 0000-0002-3609-5529); biography, roles, and awards are not independently verified.