Why did you choose to store data in JSON format instead of using Parquet, which is highly compressible?

I chose JSON for its schema‑flexibility and ease of ingestion. JSON allows me to store nested objects and arrays without pre‑defining a rigid schema, which is useful when the data source evolves or when I need to capture semi‑structured logs. It also integrates well with NoSQL stores and APIs that emit JSON. However, JSON is text‑based and has higher storage overhead compared to binary formats like Parquet. Parquet offers columnar storage, efficient compression, and faster query performance for analytical workloads. In scenarios where I need to run large‑scale analytics, I would convert JSON to Parquet during ETL. For real‑time ingestion or when the schema is volatile, JSON remains the preferred choice because it reduces the need for schema migrations and allows downstream consumers to parse only the fields they need.