Current Issue

Romanian Journal of Information Technology and Automatic Control / Vol. 34, No. 4, 2024


Programming assistance based on Naeural AI OS Platform

Tiberius PÎRCĂLABU, Nicolae ȚĂPUȘ, Andrei DAMIAN

Abstract:

In the context of the development of artificial intelligence models and particularly of the large language models (LLM), one area where they showed significant potential is programming assistance, with common tasks frequently encountered in practice, for which generic solutions already exist and can be offered as suggestions by such models. However, a complex infrastructure is required to serve an artificial intelligence model, both in terms of high-performance hardware resources and in terms of the maintenance needed for them and the related software components. This paper proposes the design and implementation of CodeXpand, an online programming assistance service by leveraging the capabilities of the Naeural AI OS, which provides a complete infrastructure for training and serving artificial intelligence models in a decentralized global community based on a token economy. This platform enables the rapid development and deployment of artificial intelligence applications by efficiently allocating infrastructure, reducing costs, and securely distributing tasks across a decentralized global community. The SDK provided by Naeural AI OS for the TypeScript language was used to integrate advanced LLM inferences for implementing a pair of tools assisting with programming tasks: a web-based conversational chat interface around programming topics and a Visual Studio Code custom extension for source code autocomplete suggestions.

Keywords:
Coding Assistant, Artificial Intelligence, Large Language Models, Blockchain, Decentralized Economy, Code Completion

View full article:

CITE THIS PAPER AS:
Tiberius PÎRCĂLABU, Nicolae ȚĂPUȘ, Andrei DAMIAN, "Programming assistance based on Naeural AI OS Platform", Romanian Journal of Information Technology and Automatic Control, ISSN 1220-1758, vol. 34(4), pp. 43-54, 2024. https://doi.org/10.33436/v34i4y202404