NDJSON / JSONL
Also known as: .ndjson file, .jsonl file, newline-delimited JSON, JSON Lines
NDJSON (newline-delimited JSON), also called JSON Lines or JSONL, stores one JSON object per line. This makes it easy to stream, append, and process large datasets record by record without loading the whole file at once.
- One JSON object per line (a.k.a. JSON Lines)
- Easy to stream, append, and process record by record
- Common in logs, ML datasets, and data pipelines
How NDJSON differs from JSON
A regular JSON file is one structure — typically an array — that you must parse in full. NDJSON instead puts a separate, complete JSON object on each line, with no surrounding array or commas between lines.
That layout means you can read a huge file one line at a time, append new records by adding lines, and split work across lines easily — ideal for logs, data exports, and streaming pipelines.
Where it is used
NDJSON/JSONL is common in data engineering, machine-learning datasets, log files, and APIs that stream results. The .ndjson and .jsonl extensions refer to the same convention.
It is plain text, so it is easy to read and grep. Files can grow large with many records, but you process them incrementally rather than all in memory.