Modern APIs ship large payloads. LLMs consume structured data token by token. Validation failures surface too late. Streaming is standard but most formats weren't designed for it. These are problems we kept running into, so we spent years working on a format built specifically for them. The core question we kept coming back to: why does every record in a dataset need to repeat the field names? The idea we landed on — separate the schema from the data entirely. Define structure once. Records contain only values. name, age, gender, joiningDt, address: {street, city, state?}, colors, isActive…

#programming #reddit #dev #data #every #record #repeat #schema