Roadmap

This page outlines the major projects that are actively in development and those planned for the near future.

In-Flight Projects

CUDA GPU Support

Adding support for GPU-accelerated compute over Vortex arrays using CUDA. This will allow encodings to be decompressed and evaluated directly on the GPU.

Lazy Array Evaluation

Reworking array compute to use lazy evaluation, where operations build up an expression graph that is materialized on demand. This is a prerequisite for efficient GPU support, since it allows the runtime to batch and schedule work across devices.

Extension DTypes

Introducing user-defined logical types that extend the built-in Vortex type system. Extension DTypes allow libraries and applications to register custom types with their own semantics while still benefiting from Vortex’s encoding, compression, and I/O infrastructure.

Upcoming Projects

Scan API

An abstract table-scan interface that positions Vortex as an interchange layer between data sources and query engines. The Scan API will support pluggable data sources and can be consumed over the C ABI for in-process integrations or over RPC for remote/distributed access.

Language Bindings Overhaul

A comprehensive rework of the language bindings to expose plugin and extension points across the ecosystem:

• Rust, C, C++, and Python will have first-class support for extending Vortex with custom encodings, compute functions, and DTypes.

• Other languages (e.g. Java) will initially focus on reading and writing Arrow data to and from Vortex files.

Tensor Extension DType

A built-in extension DType for multi-dimensional tensor data, enabling native support for fixed-shape and variable-shape tensors within Vortex arrays and files.

Variant DType

A DType for representing arbitrarily nested, JSON-like data within Vortex arrays and files. This enables efficient columnar storage and querying of semi-structured data.