Eclipse Aerospace Simplifies Flight-Data Ingestion with Databricks
Databricks, Thursday, November 12th, 2026
Eclipse Aerospace used PySpark custom data sources on Databricks to simplify ingestion of nearly 20 years of proprietary flight data.
Eclipse Aerospace, a very light jet manufacturer and operator, stores nearly 20 years of flight telemetry on Databricks, more than 3TB from over a million flight hours, in proprietary formats that need custom decoders.
Its earlier pipeline decoded files into Avro with a Rust-based decoder on AWS Lambda, which duplicated schema logic and reduced pipeline observability.
Working with Databricks, the team adopted PySpark custom data sources, available from Databricks Runtime 15.4 LTS, to read the formats directly in Spark on serverless compute.
The result is faster processing, with large flight histories analyzed in minutes, and the removal of redundant storage and multi-step processing.