Jacopo Tagliabue (Bauplan)
The self-driving data stack may be closer than it seems if we redesign OLAP systems with agents as first-class users. We identify three architectural foundations: minimal, type-safe APIs; a sandboxed, multilanguage Function-as-a-Service runtime; and Git-for-data semantics for data management. We argue that co-designing these layers offers significant advantages in safety, efficiency, and developer experience, while Apache Iceberg and clean interfaces preserve ease of integration. Drawing on our experience scaling Bauplan to millions of jobs in production, we map the emerging landscape, explain our design choices and their trade-offs, and highlight the most pressing open research problems.
Educated in several acronyms across the globe (UNISR, SFI, MIT), Jacopo is the co-founder of Bauplan, a data and AI infrastructure company. Previously, he co-founded Tooso (acquired by TSX:CVO), led Coveo's AI through its IPO, and built Coveo Labs, whose libraries, models, and datasets have garnered thousands of citations and millions of downloads. His research spans AI, NLP, information retrieval, data management, and computer systems, with collaborations at Netflix, NVIDIA, Stanford, and Wisconsin-Madison. While building Bauplan, he moonlights as a professor of ML Systems at NYU, which is notable because it’s the only job he’s ever had that his parents understand.