AI‑Microservices Portfolio

Copyleft © 2026 Gopikrishnan Rajeev
This repository contains the source for a cloud‑native, AI‑powered portfolio application.
It is built as a polyglot microservice system – the frontend runs on the DFINITY Internet Computer (ICP) while the backend services run on AWS – and is designed to demonstrate real‑world engineering practices, modern toolchains, observability, AI/ML integration, and a complete CI/CD‑driven software development lifecycle.

---

📌 Architecture Overview

High‑level components

• Frontend

  • React + Vite + TailwindCSS with GSAP animations, Three.js 3D canvas and a live AI chat widget.
  • Packaged as an ICP assets canister (frontend/dfx.json) and deployed on the Internet Computer.
  • Consent‑aware Google‑style analytics banner and visitor tracking.

• Backend microservices (Java – Spring Boot)

  • portfolio-service – CRUD API for skills, projects, work‑experience, testimonials.
  • analytics-server – visitor analytics dashboard using MongoDB.
  • notification-service – sends emails/Telegram messages.
  • api-gateway – Spring Cloud Gateway with rate‑limiting and routing.
  • config-server – centralized configuration via Spring Cloud Config.
  • All services expose /actuator/health/info/prometheus endpoints and can register with Consul.

• Python services

  • ai-service – FastAPI AI chatbot implementing Retrieval-Augmented Generation (RAG) via LangChain + OpenAI/GROQ.
    Context documents are stored as FAISS vectors with dynamic retrieval to enrich LLM replies.
    Chat history persists in PostgreSQL.
  • mcp-server – full Model Context Protocol (MCP) implementation built in FastAPI.
    It provides a secure API for other services (including the AI chatbot) to push messages into RabbitMQ, exposes Prometheus metrics, and enforces token‑based auth.

• Infrastructure & Observability

  • PostgreSQL (multiple databases), MongoDB, Redis, RabbitMQ, MinIO.
  • Prometheus node‑exporter, Grafana dashboards (see deployments/grafana-dashboards).
  • Consul for optional service discovery.
  • Docker‑Compose manifests for local development and AWS single‑instance deployment (deployments/…).
  • Alerts defined in alerts.yml.

---

🚀 Features

• AI chat widget powered by RAG-enabled LLMs; ice‑breaking questions and contextual answers are fetched from vector indexes.
• Ability for the AI (and other services) to send outbound messages via the Model Context Protocol (MCP) API.
• Visitor analytics with cookie consent banner (CookieBanner.jsx).
• Dynamic content fetching from backend services.
• CRUD operations with duplicate prevention, soft deletes.
• File uploads & presigned URLs via S3/MinIO abstraction.
• Health check library shared across Python services.
• Fluid animated UI with scroll‑triggered reveals.
• 3D WebGL background (LiquidEther.jsx) and hero island (HeroThree.jsx).

---

🛠 Technology Stack

Layer	Technologies
Frontend	React, Vite, TailwindCSS, GSAP, Three.js, react-three-fiber, Lottie
Java Backend	Java 21, Spring Boot, Spring Data JPA, Spring Cloud (Gateway, Config), Lombok, Jakarta Validation
Python APIs	Python 3.13, FastAPI, SQLAlchemy, Alembic, LangChain, FAISS, Prometheus Instrumentator
Databases	PostgreSQL, MongoDB, Redis
Messaging	RabbitMQ, MCP framework
Storage	MinIO/S3-compatible, presigned URLs
Observability	Prometheus, Grafana, Consul, psutil
Deployment	Docker Compose, NGINX, AWS EC2/cloud‑init, ICP canisters
CI/CD	GitHub Actions workflows covering build, test, lint, containerization, deployment
Misc	OpenAI/GROQ, MCP, Node/React ecosystem

---

🛠 Development & Local Setup

1. Clone repository

   git clone https://github.com/gopikrishnanrmg/Portfolio.git
   cd Portfolio

2. Infrastructure

   Add the .env file to satisfy the dependencies

   cd deployments
   docker compose --profile infra --profile services up -d

   • Spins up Postgres, Mongo, Redis, RabbitMQ, MinIO, Consul, Prometheus, Grafana, etc.
   • Environment variables come from deployments/.env.

3. Frontend

   cd ../frontend
   npm install
   npm run dev         # start dev server
   npm run build       # produce `dist` for ICP deploy

   • frontend/public/config.js is generated at runtime; inspect window.RUNTIME_CONFIG.

4. Java Services

   From each service directory (e.g. backend/portfolio-service):

   ./mvnw spring-boot:run -Dspring-boot.run.profiles=local
   # or build jar and run:
   ./mvnw clean package
   java -jar target/*.jar

5. Python Services

   cd backend/ai-service
   python -m venv .venv && source .venv/bin/activate
   pip install -r requirements.txt
   alembic upgrade head
   uvicorn app.main:app --reload

   Repeat similar steps for mcp-server.

6. Access URLs

   • Frontend: http://localhost:3000
   • AI Chat: http://localhost:8885/api/v1/chat
   • MCP Server: http://localhost:8884
   • Portfolio API (via gateway): http://localhost:8887/portfolio/api/v1/...

---

☁️ Deployment

AWS Single‑Instance

• Uses cloud‑init script in deployments/aws/singleinstance/user-data-template/user-data.
• NGINX reverse proxy configured by nginx.conf and Cloudflare tunnels.
• Services orchestrated with docker-compose.yml located in the same directory.
• Prometheus and Grafana pre‑configured with alerting rules (alerts.yml).

DFINITY Internet Computer

• Frontend assets are built and pushed to an ICP canister (frontend/dfx.json).
• Backend continues to run on AWS – the ICP host simply serves static files.

---

🧪 Testing

• Java: JUnit & Mockito unit/integration tests under src/test/java. Example: AnalyticsIntegrationTest demonstrates Testcontainers/MongoDB.
• Python: pytest can be used (tests directory not shown here).

---

🔍 Observability & Health

• Each service exposes:
  • /actuator/health, /actuator/info (Java) or equivalent FastAPI endpoints.
  • /actuator/prometheus for Prometheus scraping.
• Health checks modularised in Python (app/services/health_checks/*).
• Alerts defined in alerts.yml for Spring Boot metrics, Postgres, node CPU/disk.
• Grafana configured with dashboards (see deployments/grafana-dashboards).

---

📁 Repository Structure

.
├── backend/
│   ├── ai-service/          # Python FastAPI + LLM
│   ├── analytics-server/    # Java Spring Boot
│   ├── api-gateway/
│   ├── config-server/
│   ├── mcp-server/          # Python FastAPI MCP
│   ├── notification-service/
│   └── portfolio-service/
├── deployments/             # Docker compose, AWS templates, grafana dashboards
├── frontend/                # React + Vite UI
└── .github/                 # CI workflows

---

🎯 Why This Project?

This portfolio showcases:

• Full‑stack development across Java, Python and JavaScript.
• Microservices design with independent deployment/separation of concerns.
• AI/ML integration (vector search, LLMs, custom instructions).
• Cloud and blockchain deployment experience.
• Observability, health checks, alerting and containerised infrastructure.
• Clean code, DTOs, validation, and automated testing.
• Real‑world tooling: Docker, Prometheus, Grafana, RabbitMQ, Consul, GitHub Actions.

---

📄 License

This project is licensed under the GNU GPL v3.

---

For any questions, email gopikrishnan.rmg@outlook.com or use the chat widget on the live site.