# rares8921/enterprise-multimodal-rag-platform architecture

> Architecture diagram of the GitHub repository rares8921/enterprise-multimodal-rag-platform, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-10-03). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/rares8921/enterprise-multimodal-rag-platform
- Repository: https://github.com/rares8921/enterprise-multimodal-rag-platform

## Overview

This is a multiservice document-intelligence and RAG prototype. Its main workflow ingests tenant documents, stores metadata and raw files, processes OCR and layout asynchronously, indexes chunks, then answers authenticated queries through retrieval and LLM orchestration. Monitoring and LayoutLM fine-tuning are distinct supporting capabilities. The README documents service boundaries and workflow; source excerpts confirm several internal handoffs. Other links are shown only where the README or SOURCE INDEX supports them, so unsampled details are not inferred.

Main entry points: [`services/ingestion/main.py`](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/ingestion/main.py) (Upload API), [`services/inference-api/main.py`](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/inference-api/main.py) (Inference API), [`services/ingestion/worker.py`](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/ingestion/worker.py) (OCR worker), [`services/llm-orchestrator/main.py`](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/llm-orchestrator/main.py) (LLM orchestrator)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_ingest["Document ingestion"]
  node_ingest_api["Upload API<br/>[main.py]"]
  node_metadata[("Document metadata<br/>[models.py]")]
  node_raw_store[("Raw object storage")]
  node_queue["Task queues<br/>[task_queue.py]"]
end

subgraph group_index["Processing and indexing"]
  node_ocr_worker["OCR worker<br/>[worker.py]"]
  node_ocr_engine["EasyOCR engine<br/>[ocr_engine.py]"]
  node_layout_worker["Layout parser worker<br/>[main.py]"]
  node_layout_model["LayoutLM parser<br/>[LayoutParser.py]"]
  node_embed_worker["Embedding worker<br/>[main.py]"]
  node_embed_generator["Chunking and embeddings"]
  node_vector_store[("Tenant vector index<br/>[VectorStore.py]")]
end

subgraph group_query["Query and answers"]
  node_inference["Inference API<br/>[main.py]"]
  node_auth["Authentication and limits<br/>[dependencies.py]"]
  node_hybrid["Hybrid retrieval"]
  node_context["Context construction<br/>[context_utility.py]"]
  node_orchestrator["LLM orchestrator<br/>[main.py]"]
  node_prompt["Prompt selection<br/>[prompt_manager.py]"]
  node_complexity["Complexity scoring"]
  node_router["Model routing<br/>[ModelRouter.py]"]
  node_providers["Gemini and Mistral<br/>[GeminiLLM.py]"]
end

subgraph group_ml["Model workflow"]
  node_layout_training["LayoutLM fine-tuning"]
  node_layout_logic["Training logic<br/>[layoutlm_logic.py]"]
end

subgraph group_monitor["Monitoring"]
  node_control_plane["Monitoring control plane<br/>[app.py]"]
  node_event_store[("Monitoring event store<br/>[db.py]")]
  node_drift["Drift detection<br/>[DriftDetector.py]"]
  node_signals["RAG monitoring signals<br/>[signals.py]"]
end

node_uploader(("Document uploader"))
node_query_user(("Query user"))
node_redis[("Redis")]
node_postgres[("PostgreSQL")]
node_minio[("MinIO")]
node_pinecone[("Pinecone")]

node_uploader -->|"uploads document"| node_ingest_api
node_ingest_api -->|"writes metadata"| node_metadata
node_ingest_api -->|"stores file"| node_raw_store
node_ingest_api -->|"queues OCR"| node_queue
node_ingest_api -->|"stores status"| node_redis
node_ingest_api -->|"persists metadata"| node_postgres
node_ocr_worker -->|"dequeues work"| node_queue
node_ocr_worker -->|"reads file"| node_raw_store
node_ocr_worker -->|"runs OCR"| node_ocr_engine
node_ocr_worker -->|"writes OCR data"| node_redis
node_ocr_worker -.->|"queues layout"| node_queue
node_layout_worker -->|"reads OCR data"| node_redis
node_layout_worker -->|"parses layout"| node_layout_model
node_layout_worker -->|"stores layout"| node_redis
node_layout_worker -->|"queues embedding"| node_queue
node_embed_worker -->|"reads OCR and layout"| node_redis
node_embed_worker -->|"chunks and embeds"| node_embed_generator
node_embed_worker -->|"upserts vectors"| node_vector_store
node_vector_store -->|"stores vectors"| node_pinecone
node_query_user -->|"submits query"| node_inference
node_inference -->|"checks access"| node_auth
node_inference -->|"reranks candidates"| node_hybrid
node_inference -->|"retrieves vectors"| node_pinecone
node_inference -->|"builds context"| node_context
node_inference -->|"requests generation"| node_orchestrator
node_orchestrator -->|"selects prompt"| node_prompt
node_orchestrator -->|"scores query"| node_complexity
node_orchestrator -->|"routes model"| node_router
node_orchestrator -->|"calls provider"| node_providers
node_layout_training -->|"uses training logic"| node_layout_logic
node_control_plane -->|"writes events"| node_event_store
node_control_plane -->|"detects drift"| node_drift
node_control_plane -->|"reads signals"| node_signals

click node_ingest_api "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/ingestion/main.py"
click node_metadata "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/ingestion/models.py"
click node_queue "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/ingestion/task_queue.py"
click node_ocr_worker "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/ingestion/worker.py"
click node_ocr_engine "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/ingestion/ocr_engine.py"
click node_layout_worker "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/layout-parser/main.py"
click node_layout_model "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/layout-parser/LayoutParser.py"
click node_embed_worker "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/embedding/main.py"
click node_embed_generator "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/embedding/EmbeddingGenerator.py"
click node_vector_store "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/embedding/VectorStore.py"
click node_inference "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/inference-api/main.py"
click node_auth "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/inference-api/auth/dependencies.py"
click node_hybrid "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/inference-api/utils/hybrid_retrieval.py"
click node_context "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/inference-api/utils/context_utility.py"
click node_orchestrator "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/llm-orchestrator/main.py"
click node_prompt "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/llm-orchestrator/prompt_manager.py"
click node_complexity "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/llm-orchestrator/complexity_analyzer.py"
click node_router "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/llm-orchestrator/utils/ModelRouter.py"
click node_providers "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/llm-orchestrator/model_wrapper/GeminiLLM.py"
click node_layout_training "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/ml_pipeline/kubeflow/layoutlm_pipeline.py"
click node_layout_logic "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/ml_pipeline/kubeflow/layoutlm_logic.py"
click node_control_plane "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/monitoring/control_plane/app.py"
click node_event_store "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/monitoring/event_store/db.py"
click node_drift "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/monitoring/drift_detection/DriftDetector.py"
click node_signals "https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/monitoring/modules/rag_monitoring/signals.py"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_ingest_api,node_metadata,node_raw_store,node_queue,node_query_user toneBlue
class node_ocr_worker,node_ocr_engine,node_layout_worker,node_layout_model,node_embed_worker,node_embed_generator,node_vector_store,node_redis,node_postgres,node_minio,node_pinecone toneAmber
class node_inference,node_auth,node_hybrid,node_context,node_orchestrator,node_prompt,node_complexity,node_router,node_providers toneMint
class node_layout_training,node_layout_logic toneRose
class node_control_plane,node_event_store,node_drift,node_signals,node_uploader toneIndigo
```

## Components

### Document ingestion

- [Upload API](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/ingestion/main.py): `services/ingestion/main.py`
- [Document metadata](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/ingestion/models.py): `services/ingestion/models.py`
- **Raw object storage**
- [Task queues](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/ingestion/task_queue.py): `services/ingestion/task_queue.py`

### Processing and indexing

- [OCR worker](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/ingestion/worker.py): `services/ingestion/worker.py`
- [EasyOCR engine](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/ingestion/ocr_engine.py): `services/ingestion/ocr_engine.py`
- [Layout parser worker](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/layout-parser/main.py): `services/layout-parser/main.py`
- [LayoutLM parser](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/layout-parser/LayoutParser.py): `services/layout-parser/LayoutParser.py`
- [Embedding worker](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/embedding/main.py): `services/embedding/main.py`
- [Chunking and embeddings](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/embedding/EmbeddingGenerator.py): `services/embedding/EmbeddingGenerator.py`
- [Tenant vector index](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/embedding/VectorStore.py): `services/embedding/VectorStore.py`

### Query and answers

- [Inference API](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/inference-api/main.py): `services/inference-api/main.py`
- [Authentication and limits](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/inference-api/auth/dependencies.py): `services/inference-api/auth/dependencies.py`
- [Hybrid retrieval](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/inference-api/utils/hybrid_retrieval.py): `services/inference-api/utils/hybrid_retrieval.py`
- [Context construction](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/inference-api/utils/context_utility.py): `services/inference-api/utils/context_utility.py`
- [LLM orchestrator](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/llm-orchestrator/main.py): `services/llm-orchestrator/main.py`
- [Prompt selection](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/llm-orchestrator/prompt_manager.py): `services/llm-orchestrator/prompt_manager.py`
- [Complexity scoring](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/llm-orchestrator/complexity_analyzer.py): `services/llm-orchestrator/complexity_analyzer.py`
- [Model routing](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/llm-orchestrator/utils/ModelRouter.py): `services/llm-orchestrator/utils/ModelRouter.py`
- [Gemini and Mistral](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/llm-orchestrator/model_wrapper/GeminiLLM.py): `services/llm-orchestrator/model_wrapper/GeminiLLM.py`

### Model workflow

- [LayoutLM fine-tuning](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/ml_pipeline/kubeflow/layoutlm_pipeline.py): `ml_pipeline/kubeflow/layoutlm_pipeline.py`
- [Training logic](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/ml_pipeline/kubeflow/layoutlm_logic.py): `ml_pipeline/kubeflow/layoutlm_logic.py`

### Monitoring

- [Monitoring control plane](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/monitoring/control_plane/app.py): `services/monitoring/control_plane/app.py`
- [Monitoring event store](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/monitoring/event_store/db.py): `services/monitoring/event_store/db.py`
- [Drift detection](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/monitoring/drift_detection/DriftDetector.py): `services/monitoring/drift_detection/DriftDetector.py`
- [RAG monitoring signals](https://github.com/rares8921/enterprise-multimodal-rag-platform/blob/main/services/monitoring/modules/rag_monitoring/signals.py): `services/monitoring/modules/rag_monitoring/signals.py`

### Other components

- **Document uploader**
- **Query user**
- **Redis**
- **PostgreSQL**
- **MinIO**
- **Pinecone**

## Connections

- Document uploader → Upload API: uploads document
- Upload API → Document metadata: writes metadata (evidence: `services/ingestion/main.py`)
- Upload API → Raw object storage: stores file
- Upload API → Task queues: queues OCR (evidence: `services/ingestion/main.py`)
- Upload API → Redis: stores status (evidence: `services/ingestion/main.py`)
- Upload API → PostgreSQL: persists metadata (evidence: `services/ingestion/main.py`)
- OCR worker → Task queues: dequeues work (evidence: `services/ingestion/worker.py`)
- OCR worker → Raw object storage: reads file (evidence: `services/ingestion/worker.py`)
- OCR worker → EasyOCR engine: runs OCR (evidence: `services/ingestion/worker.py`)
- OCR worker → Redis: writes OCR data
- OCR worker → Task queues: queues layout
- Layout parser worker → Redis: reads OCR data (evidence: `services/layout-parser/main.py`)
- Layout parser worker → LayoutLM parser: parses layout (evidence: `services/layout-parser/main.py`)
- Layout parser worker → Redis: stores layout (evidence: `services/layout-parser/main.py`)
- Layout parser worker → Task queues: queues embedding (evidence: `services/layout-parser/main.py`)
- Embedding worker → Redis: reads OCR and layout (evidence: `services/embedding/main.py`)
- Embedding worker → Chunking and embeddings: chunks and embeds (evidence: `services/embedding/main.py`)
- Embedding worker → Tenant vector index: upserts vectors (evidence: `services/embedding/main.py`)
- Tenant vector index → Pinecone: stores vectors
- Query user → Inference API: submits query
- Inference API → Authentication and limits: checks access (evidence: `services/inference-api/main.py`)
- Inference API → Hybrid retrieval: reranks candidates (evidence: `services/inference-api/main.py`)
- Inference API → Pinecone: retrieves vectors
- Inference API → Context construction: builds context
- Inference API → LLM orchestrator: requests generation
- LLM orchestrator → Prompt selection: selects prompt (evidence: `services/llm-orchestrator/main.py`)
- LLM orchestrator → Complexity scoring: scores query (evidence: `services/llm-orchestrator/main.py`)
- LLM orchestrator → Model routing: routes model (evidence: `services/llm-orchestrator/main.py`)
- LLM orchestrator → Gemini and Mistral: calls provider (evidence: `services/llm-orchestrator/main.py`)
- LayoutLM fine-tuning → Training logic: uses training logic (evidence: `ml_pipeline/kubeflow/layoutlm_pipeline.py`)
- Monitoring control plane → Monitoring event store: writes events (evidence: `services/monitoring/control_plane/app.py`)
- Monitoring control plane → Drift detection: detects drift (evidence: `services/monitoring/control_plane/app.py`)
- Monitoring control plane → RAG monitoring signals: reads signals (evidence: `services/monitoring/control_plane/app.py`)
