I build large-scale historical datasets from archival image scans using multimodal large language models to study how individuals, firms, and cities interacted with the German patent system and responded to technological change during the Second Industrial Revolution.
Alongside this work, I develop AI Co-Historians in AI for History (AI4History) to support historical research workflows and help historians analyse millions of previously unread primary sources. My goal is to accelerate historical research while giving historians more agency and leverage in understanding our past.
Papers
Joint work with equal contribution.
Chronos: The AI Co-Historian
with Lorenz Hufe, Gavin Greif, Sebastian Oliver Eck, and Philip Torr
Multimodal LLMs for Historical Dataset Construction from Archival Image Scans: German Patents (1877-1918)
with Jochen Streb
Multimodal LLMs for OCR, OCR Post-Correction, and NER in Historical Documents
with Gavin Greif and Robin Greif