跨语言的列式内存格式与工具箱，让不同系统之间零拷贝交换数据，并加速内存分析。
面向多模态 AI 的开放湖仓数据格式，两行代码即可从 Parquet 转换，实现百倍随机访问加速、向量索引和数据版本管理，兼容 Pandas、DuckDB、PyTorch 等生态。
面向 AI 与多模态负载的高性能数据引擎，用 Rust 编写，能大规模处理图像、音频、视频和结构化数据，适合搭建 ETL 与数据管道。
用 Rust 构建的统一基础设施可观测平台，基于数据湖架构统一采集日志、指标、链路和事件，靠列式遥测压缩把存储成本最多降低 90%。
命令行工具，用于在 3DGS、KSplat、SPZ、Splat 等十余种 3D 高斯泼溅模型格式间互转，并支持 GPU 加速的统计离群点与密度滤波。
基于 DuckDB 的 TypeScript 数据分析库，能快速处理表格、地理空间、图网络和向量相似度检索，还支持 AI 与 Google Sheets 集成。
…ub.com/datachain-ai/datachain/pull/2161 fix: read Parquet map columns as dicts by @shcheklein in https://github.com/datachain-ai/datachain/pull/2162 fix: read values with
…ithub.com/weaviate/weaviate/pull/13592 deps: bump parquet-go/parquet-go to v0.32.0 and andybalholm/brotli to v1.2.7 by @antas-marcin in https://github.com/weaviate/weavia
…ithub.com/weaviate/weaviate/pull/13632 deps: bump parquet-go/parquet-go to v0.32.0 and andybalholm/brotli to v1.2.7 by @antas-marcin in https://github.com/weaviate/weavia
…//github.com/datachain-ai/datachain/pull/2130 fix(parquet): read only the picked columns with ParquetFileFormat by @shcheklein in https://github.com/datachain-ai/datachai
…failing to read relative local paths: read remote parquet now resolves relative local paths to absolute before passing them to Ray Data. Ray 2.56+ schedules read tasks on
…-prep-kit/data-prep-kit/pull/1528 updated docling2parquet to remove image data from contents field when… by @swith005 in https://github.com/data-prep-kit/data-prep-kit/pu