Skip to content

This is an legacy deployment system implementation for PureCPP an open-source modular RAG (Retrieval-Augmented Generation) system developed by PureAI,. My contributions include C++<->Python bindings, vector database, modular chunks, and deploy scripts. Initial design evolved into the backbone architecture. Published under MIT license

License

Notifications You must be signed in to change notification settings

bbzaffari/Deploy-System

Repository files navigation

PureCPP

This is a legacy deployment system implementation — before moving to GitHub Actions — for PureCPP an open-source modular RAG (Retrieval-Augmented Generation) system developed by PureAI, to which I contributed during my internship, integrating C++ and Python, deploying to PyPI, and optimizing for high-performance vector processing and queries. It also includes careful memory control strategies to maximize performance, culminating in the development of a functional and scalable vector database engine. In addition to its modular architecture, the project features orchestration and optimization through build and deploy pipeline scripts, reducing processes that originally took ~3 hours down to just 30 minutes.


Note:
As this project is extensive and contains many modular components, this documentation will initially focus on explaining the parts I developed in the about the deploy system, as well as their integration with the pip

Later sections will extend the documentation to cover:


About

This is an legacy deployment system implementation for PureCPP an open-source modular RAG (Retrieval-Augmented Generation) system developed by PureAI,. My contributions include C++<->Python bindings, vector database, modular chunks, and deploy scripts. Initial design evolved into the backbone architecture. Published under MIT license

Topics

Resources

License

Stars

Watchers

Forks