aperiodic-io/tparquet: Parquet for time series data compression using pcodec

Tparquet is a new lossless compression codec extension for Apache Parquet that integrates the pcodec library to significantly reduce the size of numeric time-series data files. Tested across multiple market data providers, the format achieves about half the size of standard ZSTD-compressed Parquet files while preserving schemas, row groups, and statistics. The project provides complete cross-language support and bindings across Python, Rust, Go, Java, DuckDB, Spark, and Trino. Open article →