About
I’m Ricardo Ledan — an AI systems engineer and independent researcher working across artificial intelligence, knowledge systems, and digital archives.
I run studio1804, an independent AI systems practice, and created rasin.ai, a source-grounded retrieval and knowledge graph platform for Haitian historical research. Rasin.ai is one expression of a broader question that drives my work: how should technical systems represent knowledge when the evidence is incomplete, multilingual, politically shaped, and contested?
My work focuses on the systems layer of AI: knowledge graphs for archives, provenance-aware retrieval, multi-agent document analysis, and self-hosted AI infrastructure that does not depend entirely on the cloud. I’m interested in how knowledge is structured, how models interact with evidence over time, and what makes these systems reliable, interpretable, and useful outside controlled environments.
A lot of this comes back to questions of preservation, ownership, access, and trust. Who owns the infrastructure public knowledge depends on? How do technical systems shape what people are able to find and understand? And how can public-facing AI systems preserve uncertainty, disagreement, and provenance instead of flattening them?
I write here about the systems I’m building and the ideas behind them — archival AI, retrieval systems, agent infrastructure, digital humanities, historical archives, and questions of access, provenance, and trust.
Reading
- What Is Intelligence? by Blaise Agüera y Arcas — on intelligence as prediction across biological and artificial systems.
- Machine Decision Is Not Final, edited by Benjamin H. Bratton — on AI, politics, cybernetics, and decision-making beyond the Silicon Valley frame.
- Build a Large Language Model (From Scratch) by Sebastian Raschka — a technical walkthrough of the model stack from tokenization through fine-tuning.