mia-platform/ai-rag-template

How mia-platform/ai-rag-template works

This backend serves RAG chat and document ingestion. Chat requests enter the API, retrieve similar MongoDB Atlas documents, and pass query, history, and context through the assistant chain to an LLM; ingestion accepts URLs or uploaded files, parses and chunks text, then writes embeddings to the vector store. It also exposes operational health and Prometheus metrics, and updates the vector index at startup. The README documents these capabilities; some implementation details are unsampled, so their internal links are omitted.

Main entry points: app.py FastAPI application · assistant_service.py Assistant service · embedding_service.py Embedding service · embeddings_handler.py Embedding endpoints

Components 27

API and operations

FastAPI application src/app.py
Chat endpoint src/api/controllers/chat_completions/chat_completions_handler.py
Embedding endpoints src/api/controllers/embeddings/embeddings_handler.py
Health endpoints src/api/controllers/core
Metrics endpoint src/api/controllers/core/metrics/metrics_handler.py
Request context src/api/middlewares/app_context_middleware.py

RAG chat

Assistant service src/application/assistant/assistant_service.py
Assistant chain src/application/assistant/chains/assistant_chain.py
Prompt builder src/application/assistant/chains/assistant_prompt.py
Document combine chain src/application/assistant/chains/combine_docs_chain.py
Vector retriever src/application/assistant/chains/retriever_chain.py

Document ingestion

Embedding service src/application/embeddings/embedding_service.py
File parser src/application/embeddings/file_parser/file_parser.py
Document chunker src/application/embeddings/document_chunker.py
Hyperlink parser src/application/embeddings/hyperlink_parser.py
Vector index updater src/lib/vector_search_index_updater.py

Models and infrastructure

Application context src/context.py
Service configuration src/configurations/service_model.py
Configuration loader src/configurations/configuration.py
Embedding provider src/infrastracture/embeddings_manager/embeddings_manager.py
LLM provider src/infrastracture/llm_manager/llm_manager.py
Prometheus metrics src/infrastracture/metrics_manager/metrics_manager.py

Other components

Chat or ingest client
MongoDB Atlas vector store
LLM API
Website
Uploaded documents

Connections 32

Chat or ingest client → FastAPI application: sends requests README.md
FastAPI application → Chat endpoint: registers route src/app.py
FastAPI application → Embedding endpoints: registers route src/app.py
FastAPI application → Health endpoints: registers routes src/app.py
FastAPI application → Metrics endpoint: registers route src/app.py
FastAPI application → Request context: installs src/app.py
Chat endpoint → Assistant service: invokes src/api/controllers/chat_completions/chat_completions_handler.py
Assistant service → Assistant chain: invokes src/application/assistant/assistant_service.py
Assistant service → Vector retriever: configures src/application/assistant/assistant_service.py
Assistant service → Prompt builder: builds prompt src/application/assistant/assistant_service.py
Assistant service → Document combine chain: configures src/application/assistant/assistant_service.py
Assistant service → LLM provider: creates model src/application/assistant/assistant_service.py
Assistant chain → Vector retriever: retrieves documents src/application/assistant/chains/assistant_chain.py
Vector retriever → MongoDB Atlas vector store: similarity searches src/application/assistant/chains/retriever_chain.py
Assistant service → MongoDB Atlas vector store: returns references README.md
LLM provider → LLM API: calls model
Embedding endpoints → Embedding service: starts generation src/api/controllers/embeddings/embeddings_handler.py
Embedding endpoints → File parser: parses upload src/api/controllers/embeddings/embeddings_handler.py
Embedding service → Document chunker: chunks text src/application/embeddings/embedding_service.py
Embedding service → Hyperlink parser: extracts links src/application/embeddings/embedding_service.py
Embedding service → Embedding provider: creates embeddings src/application/embeddings/embedding_service.py
Embedding provider → MongoDB Atlas vector store: writes vectors
Embedding service → MongoDB Atlas vector store: stores documents src/application/embeddings/embedding_service.py
Website → Embedding service: provides pages README.md
Uploaded documents → Embedding endpoints: uploads content README.md
FastAPI application → Vector index updater: updates index src/app.py
Vector index updater → MongoDB Atlas vector store: creates or updates index README.md
FastAPI application → Configuration loader: loads configuration src/app.py
Configuration loader → Service configuration: validates configuration src/configurations/configuration.py
FastAPI application → Application context: creates context src/app.py
Request context → Application context: provides request context src/api/middlewares/app_context_middleware.py
Metrics endpoint → Prometheus metrics: exposes metrics src/api/controllers/core/metrics/metrics_handler.py

More Python diagrams: letterligo/agent-webcloak · lukasroom/superadd · marketcalls/agentic-trader · mogvision/hvit · moritzhambach/detecting-cheating-in-chess

Drawn by GitDiagram on September 30, 2026 from the repository's file tree, README and source files. Read as Markdown ·