What happens when a researcher turns his own CV into a dataset

A CV is designed to be read, not analyzed. It lists what you did in reverse chronological order and trusts the reader to form an impression. Before coming to New York to study data visualization, I spent nearly a decade building an academic career in Russia — journal articles, co-authored textbooks, a doctoral dissertation. For this project I took that list, turned it into a proper dataset, and let the data describe the career instead of the narrative.

Building the dataset

The dataset has 30 observations — every publication I authored or co-authored between 2018 and 2023 — assembled from my CV and cross-referenced against eLIBRARY, Russia’s main academic database, and the archive of the journal where nine of my articles appeared. For each publication I coded six variables: year, venue, type, topic, methodology, and co-authorship. On top of that, I ran a word-frequency analysis across the 20 publications whose full texts were digitally available: extracting text, removing stop words, translating terms to English, and counting occurrences by year.

One methodological note worth making explicitly: topic categories were assigned by me, and that introduces subjectivity. A paper about burnout in orphanage workers could sit under “Burnout & Wellbeing” or “Labor Motivation.” I coded each work by its central argument — and flagging that choice is part of doing this honestly. Coding decisions are analysis decisions, whether the dataset describes customers, employees, or yourself.

What the dashboards showed

Seventy percent of the publications were solo-authored. Output peaked at nine publications in 2022, at the height of the dissertation cycle, and the composition shifted visibly over time: solo empirical work dominated the early years, while co-authored textbooks on digital sociology and urban development emerged from 2019 onward. Measured in pages rather than counts, my dissertation topic — labor motivation — dominates with over 540 pages. But the co-authored page count clusters around broader teaching topics, a reminder that page volume is a poor proxy for research effort: a seven-page article can carry more original work than a 144-page textbook.

By type, journal articles make up just over half the output (53%), with textbooks and conference papers at 17% each — a typical profile for an early-career researcher in the Russian academic system. By methodology, 43% of the works were quantitative, 33% theoretical, and 23% mixed. The arc is visible in the data: theoretical papers early, quantitative work rising as I collected original survey data for the dissertation.

The methodology mix tells the same story from another angle.

The word-frequency analysis confirmed the career transition more sharply than memory would. “Social” leads across the corpus (1,282 occurrences), followed by “development” (640). “Motivation” and “labor” peak around 2020 and decline; “digital,” “urban,” and “technology” surge in 2022 with the textbook wave. The dataset caught a professional pivot I had lived through without consciously registering.

Why this matters beyond one CV

The limitations are the honest part of the exercise: the text analysis covers 20 of 30 works, because some existed only as scanned images without a text layer; the topic coding is one researcher’s judgment; page count measures length, not value. Every dataset a consultant touches has equivalents of these three problems — coverage gaps, coder subjectivity, misleading proxies — and the discipline is the same: name them, bound them, and interpret inside those bounds.

Data is never neutral. Even a list of publications carries the fingerprints of the person who created it — the choices about what to study, whom to collaborate with, where to publish. Turning my CV into a dataset did not just quantify a decade; it made me see it differently. That, in miniature, is what a good analysis does for an organization: not new facts so much as a new angle on facts it already owns.