Keymakr joins RUKOPYS as labeling partner for Ukrainian handwriting dataset

By AI, Created 9:30 AM UTC, May 25, 2026, /AGP/ – Keymakr has become the official labeling partner for RUKOPYS, a new open dataset of Ukrainian handwritten text spanning more than 100 years. The project is designed to help train OCR and handwriting recognition models, support government digitization efforts, and preserve Ukrainian language and archival records.

Why it matters: - RUKOPYS fills a major gap in Ukrainian AI development by creating a large open dataset for handwriting recognition. - The dataset can help train OCR and HTR models on real-world handwriting, including archival material and modern notebooks. - The project is meant to support government digitization, improve access to archival data, and preserve Ukrainian language and historical records.

What happened: - Keymakr became the official labeling partner for RUKOPYS, described as the first comprehensive annotated dataset of Ukrainian handwritten text. - The dataset covers more than 100 years of Ukrainian handwriting, from archived documents from the 1920s to modern notebooks. - RUKOPYS was initiated by the Ministry of Economy, Environment, and Agriculture of Ukraine, with support from the Ministry of Digital Transformation. - The project was developed in collaboration with the Ukrainian Catholic University and AI HOUSE, a Ukrainian nonprofit focused on growing the country’s AI community.

The details: - RUKOPYS combines materials collected with national institutions, universities, and archival organizations. - The dataset is fully anonymized and available under an open license for research and education. - Keymakr supported the dataset creation process through manual annotation by in-house experts. - The work involved line-level structuring, transcription, normalization, and quality control for difficult handwritten records. - The project had to account for changing scripts, degraded archival materials, low legibility, spacing inconsistencies, and formatting differences across decades. - Zoya Boyko, PM at Keymakr, said the team validated each text line, refined edge cases with AI HOUSE, and proposed workflow optimizations that improved efficiency and data quality. - Boyko said the initial expectations were exceeded and the result was a valuable dataset for Ukrainian model training. - The dataset is the foundation for the Handwritten to Data AI Challenge, an open competition focused on recognizing applications, certificates, logs, signatures, stamps, and archival documents. - The challenge is intended to identify a solution for possible integration into the ePermit system. - AI/ML engineers, data scientists, researchers, students, startups, R&D teams, and university labs gained access to RUKOPYS. - Participants also received access to Amazon Web Services for model training and the chance to test solutions in practical scenarios with potential government use.

Between the lines: - The project reflects a broader push to turn paper archives into structured data that can be reused in public infrastructure and AI development. - Dmytro Voitekh, AI advisor to the Ministry of Economy of Ukraine and AI/ML Lead at Mriya, said public datasets can improve Ukrainian-language model quality and volume without major extra costs. - Voitekh said competition platforms such as Kaggle can help surface models that could support future digitization pipelines, archival analysis, and government services. - Keymakr framed its role as part of a larger effort to build reliable training data pipelines for difficult AI use cases and socially useful projects. - Inna Nomerovska, Chief Marketing Officer at Keymakr, said the company sees the effort as a contribution to document digitization and to preserving Ukrainian culture and language. - Nomerovska noted that Ukraine ranks 5th globally in digital public services and wants to be among the top three countries in public-sector AI use by 2030.

What’s next: - The Handwritten to Data AI Challenge will test competing Computer Vision solutions against real document-processing tasks. - Organizers aim to select the best-performing approach for possible integration into ePermit and other government workflows. - The broader goal is to speed up document processing, improve accessibility to archival material, and advance digital transformation in public institutions. - More information is available in Keymakr’s announcement and on the company’s YouTube channel.

The bottom line: - Keymakr’s role in RUKOPYS gives Ukraine a stronger open foundation for handwriting AI while tying technical progress to public-sector digitization and cultural preservation.