Noeon Research is a startup conducting research on the architecture that is an alternative to Transformers-based ones. The company’s goal is to build an AI system for automated code processing for complex IT projects.
Noeon Research focuses on the automatisation of code synthesis and analysis using a hybrid approach. Our solution incorporates the ability to process a large amount of data from Machine Learning and the ability to have a strong inductive bias by efficient prior knowledge encoding.
Architecture has these properties
Developer is able to follow the AI system’s reasoning and examine the internal logic. This is essential for the system’s safety and reliability.
The cost of incorporating new domain specific knowledge into a project is the same at both the initial and later stages. This quality enables fast project requirement updates.
The system is able to identify a lack of information and ask a developer for clarifications. This accelerates project development by significantly decreasing feedback time.
The system does not lose any context, which makes work on complex IT projects possible.