Upload a PDF or JSONL file and we will queue ingestion, chunk with overlap, and store embeddings in Postgres.
For JSONL uploads, each line is treated as one chunk and chunk settings are ignored.