An independent AI research lab building multilingual systems for low-resource languages and the archives that hold them.
Flagship Project
Rasin makes the archive searchable.
256,000+ pages, four languages, 500 years of records — every answer cited to the exact document and page. Colonial records, revolutionary dispatches, constitutional texts, newspapers: indexed across French, Kreyòl, English, and Spanish, with historiography intact.
Detail — archival source, Rasin corpus
Haitian Kreyòl + Multilingual AI
Rasin tests multilingual OCR, embeddings, and retrieval on Haitian Kreyòl alongside French, English, and Spanish across 256,000+ historical pages.
Archives as Research Instruments
Rasin turns scattered primary sources into a searchable, citable research surface where every answer points back to the exact document and page.
Published, Citable Methods
We document the pipeline behind the work: OCR, entity extraction, retrieval, source citation, and the security patterns needed for AI systems that handle external tools.
Partnerships for Funders, Scholars, and Archives
Three clear doors into the lab: fund archive-scale work, co-author technical research, or adapt multilingual search systems for cultural collections.
Haitian Kreyòl + Multilingual AI
Rasin tests multilingual OCR, embeddings, and retrieval on Haitian Kreyòl alongside French, English, and Spanish across 256,000+ historical pages.
Explore ›Archives as Research Instruments
Rasin turns scattered primary sources into a searchable, citable research surface where every answer points back to the exact document and page.
Explore ›Published, Citable Methods
We document the pipeline behind the work: OCR, entity extraction, retrieval, source citation, and the security patterns needed for AI systems that handle external tools.
Explore ›Partnerships for Funders, Scholars, and Archives
Three clear doors into the lab: fund archive-scale work, co-author technical research, or adapt multilingual search systems for cultural collections.
Explore ›Membership
studio1804 is a member of the NVIDIA Inception program.
NVIDIA Inception gives studio1804 access to GPU-accelerated compute for the work Rasin requires — OCR at scale, multilingual retrieval across four languages, and entity indexing across 256,000+ pages of historical documents.
