Pometry is architected from the ground up to treat time as a fundamental aspect of data. Every piece of data, every relationship, and every property is understood within its temporal context, allowing for efficient queries and analytics that track evolution and change as first-class citizens, rather than an afterthought.

Through a combination of highly optimised Rust code, advanced techniques like lock-free data structures, vectorised execution (leveraging SIMD/AVX512), efficient memory management, and an architecture designed specifically for graph workloads. This allows us to maximise the utility of modern CPUs without requiring massive clusters for many demanding tasks.

Pometry is extremely efficient where others are not. We employ advanced compression techniques that significantly reduce the storage footprint of your data, temporal or static. We’ve demonstrated compressing 4TB of raw CSV data down to just 350GB in our optimized graph format. With our highly efficient on-fly decompression, we ensure you are able to save on storage cost without performance impact on analytics. This means you can handle very large datasets without necessarily needing massive storage infrastructure, contributing to lower costs and better performance on commodity hardware.

The small footprint enables incredible deployment flexibility: run Pometry on edge devices, embed it directly within your Python ML pipelines for zero-latency analytics, deploy in resource-constrained environments, or use it as a lightweight, powerful server. It simplifies distribution and reduces overhead.

Pometry is designed to be open. It can ingest data from virtually any source (CSVs, Parquet, Kafka, SQL databases, data lakes like S3/Azure Blob) and write results and enriched data back out to any system. It doesn't require you to centralise all your data within it.

Python APIs offer ease of use, rapid development, and seamless integration with the rich Python data science ecosystem. The high-performance Rust APIs are for advanced users who need to extract every ounce of performance, work directly with data buffers, or build custom, ultra-low-latency extensions using lock-free mechanisms. The high-level Rust and Python APIs hold a 1:1 parity, with the HPC Rust APIs being slightly more complex tailored for advanced users.

Pometry's ability to ingest from and write to any data source, its small footprint allowing for flexible deployment (including ephemeral use), and its design philosophy of not needing to be the "centre of the universe" mean you can integrate it strategically without being forced into a proprietary ecosystem for all your data needs.