AI & MLOps Flagship Platform Hub

An integrated flagship MLOps platform combining GPU telemetry monitoring, model serving registries, LangChain RAG, edge video inference, sandboxed code execution, and LLM benchmarking.

This monorepo aggregates multiple advanced standalone blueprints into a unified multi-service ecosystem. It represents an enterprise-grade approach where services share common infrastructure while remaining modular and independently deployable.

---

🗺️ Shared Architecture Diagram

flowchart TD
    USER[User / Developer] --> WEB[Next.js Gateway Console]
    WEB --> services[Individual Microservices]
    
    subgraph Services [Integrated Services Workspace]
        service_1["AI Cluster & Local LLM Telemetry Dashboard"]
        service_5["Edge AI Video Analytics Control Plane"]
        service_6["Multi-Tenant RAG Knowledge Platform"]
        service_7["Kubernetes GPU MLOps Studio"]
        service_33["JupyterHub Multi-User GPU Lab Portal"]
        service_34["CUDA Benchmark Lab & Model Serving Registry"]
        service_35["Local LLM Evaluation Arena"]
        service_44["Library Management & Recommendation Engine"]
        service_50["AI Coding Lab & Sandbox Evaluation Platform"]
    end
    
    services --> Services
    
    subgraph Infrastructure [Shared Services Infrastructure]
        DB[(Shared PostgreSQL)]
        CACHE[(Shared Redis)]
        MQ[(Shared MQTT Broker)]
      end
      
    Services --> DB
    Services --> CACHE
    Services --> MQ

Loading

---

📂 Sub-Services Index

This platform contains the following sub-services (located in the services/ folder):

• services/01-ai-cluster-llm-telemetry - AI Cluster & Local LLM Telemetry Dashboard

• services/05-edge-ai-video-analytics-control-plane - Edge AI Video Analytics Control Plane
  Monitor camera streams, run edge inference, review detections, and manage model rollout across devices.

• services/06-multi-tenant-rag-knowledge-platform - Multi-Tenant RAG Knowledge Platform
  A secure knowledge base that ingests documents, chunks content, embeds text, and serves tenant-isolated RAG answers.

• services/07-kubernetes-gpu-mlops-studio - Kubernetes GPU MLOps Studio
  Schedule GPU jobs, track experiment metadata, monitor pods, and compare training runs from a web console.

• services/33-jupyterhub-gpu-lab-portal - JupyterHub Multi-User GPU Lab Portal
  Provision user notebooks, allocate GPU access, track usage, and manage classroom/lab environments.

• services/34-cuda-benchmark-model-registry - CUDA Benchmark Lab & Model Serving Registry
  Benchmark GPU readiness, model latency, memory pressure, and serving configurations across machines.

• services/35-local-llm-evaluation-arena - Local LLM Evaluation Arena
  Compare local models with prompt suites, scoring rubrics, latency metrics, and side-by-side outputs.

• services/44-library-recommendation-management-system - Library Management & Recommendation Engine
  Manage books, borrowers, circulation, fines, and AI-powered book recommendations.

• services/50-ai-code-sandbox-evaluation-platform - AI Coding Lab & Sandbox Evaluation Platform
  Run code safely, evaluate AI-generated solutions, score tests, and display execution traces in a lab UI.

---

⚡ Quick Start

1. Boot Shared Infrastructure

Spin up the shared Postgres database and Redis cache from the root directory:

docker compose up -d

2. Run a Specific Service

To run any of the integrated services:

cd services/<service-folder-name>
# Copy environment variables
cp .env.example .env
# Boot the service container stack
docker compose up --build

---

🧑‍💻 Workspace Management

This monorepo uses NPM workspaces for package sharing.

• Install all dependencies: npm install
• Run web apps in dev mode: npm run dev:web (from workspaces root)

Designed and built by Yash Pradip Khaire.