How chirayuagrawal-dev/cloud-native-microservice-pipeline works
Aurora AI lets a user ask a question through the frontend; the Express API validates it, retrieves matching document chunks from ChromaDB, builds a context-limited prompt, and asks Ollama to generate an answer. A separate ingestion workflow loads project and folder documents, chunks them, and indexes them in the same collection using the shared embedding function. The README also documents Kubernetes workloads for these services. Frontend and deployment details are not sampled here, so those links rely on README statements; no CI/CD internals are included.
Main entry points: k8s-deployment.yaml Kubernetes workloads · ingest.js Collection indexing · server.js Context retrieval · frontend Aurora frontend
Components 14
User experience
- Aurora frontend
frontend
RAG API
- Question API
server.js - Question validation
server.js - Context retrieval
server.js - Prompt assembly
server.js - Embedding function
embeddings.js
Document ingestion
- Source documents
documents - Document loading
ingest.js - Chunk preparation
ingest.js - Collection indexing
ingest.js
AI and deployment runtime
- ChromaDB
- Ollama
- Kubernetes workloads
k8s-deployment.yaml
Other components
- User
Connections 21
- User → Aurora frontend: asks question
README.md - Aurora frontend → Question API: posts question
README.md - Question API → Question validation: validates
server.js - Question validation → Context retrieval: queries
server.js - Context retrieval → ChromaDB: queries collection
server.js - Context retrieval → Embedding function: supplies function
server.js - Embedding function → Ollama: creates embeddings
embeddings.js - Context retrieval → Prompt assembly: provides context
server.js - Prompt assembly → Ollama: requests generation
server.js - Question API → Aurora frontend: returns answer
README.md - Source documents → Document loading: loads files
ingest.js - Document loading → Chunk preparation: prepares text
ingest.js - Chunk preparation → Collection indexing: provides chunks
ingest.js - Collection indexing → Embedding function: configures function
ingest.js - Collection indexing → Ollama: checks model
ingest.js - Collection indexing → ChromaDB: writes collection
ingest.js - Kubernetes workloads → Aurora frontend: runs workload
README.md - Kubernetes workloads → Question API: runs workload
README.md - Kubernetes workloads → ChromaDB: runs workload
README.md - Kubernetes workloads → Ollama: runs workload
README.md - Kubernetes workloads → Collection indexing: runs job
README.md