I build open-source document/OCR tools in TypeScript. Right now I’m working on OCRBase: a self-hostable, model-agnostic API that turns PDFs and images into Markdown or validated JSON, plus a small typed SDK around it. Currently wrestling with async jobs, queues, object storage, and making VLM-backed extraction reliable in production.
Type-safe TypeScript SDK for docling-serve with first-class Bun support. The client is generated from the official OpenAPI schema and wraps every endpoint with a small, DX-focused API surface.