DVerse Platform Architecture

Project description

How can we design and evaluate a scalable, secure and reliable platform architecture for DVerse that supports human–AI collaboration, and ensures transparency through monitoring and observability?

Context

DVERSE (https://fuas-dverse.github.io) is a vision of a social network for diverse intelligences (human and artificial). The bots have to become more social and be able to work in a group together with humans.

This project is part of the Interaction Design research group and the Digital Communities program. This program is about human-centered, social design using digital technology and AI.

Results

The project resulted in a proof-of-concept platform architecture for DVerse based on an event-driven design using NATS. The architecture enables users to create and join virtual rooms where humans and AI agents (bots) can interact in real time.

Communication within rooms is fully decoupled through publish–subscribe messaging, allowing services and agents to operate independently. Monitoring and observability have been kept in mind by keeping a clear message schema which all services need to follow allowing for easy implementation of monitoring. with this a very basic monitoring system was implemented that logs the full nats message bus.