Part 5/13:
He describes his process of merging microservices, especially combining the "embedding" microservice with the main "Nexus" service. This consolidation reduces unnecessary complexity, as storing embeddings locally within Nexus simplifies data flow and reduces inter-service dependencies. The rationale is straightforward: if a small, one-use microservice exists only in one place, it may not warrant the overhead of being separated.
To illustrate this, he shows how the Nexus logs data entries—including content, microservice origin, timestamps, UUIDs, and vector embeddings—in JSON format for readability, explanation, and experimentation. He admits that storing large vectors (such as 512-dimensional embeddings) as JSON lists is not scalable long-term, but suits early prototypes well.