WealthEngine – Build Walkthrough

What Was Built

WealthEngine is a 3-tier financial analysis platform for Indian retail investors.

---

Architecture Overview

graph TD
    A[Angular 19 Frontend] -->|JWT REST| B[api-gateway :8080]
    B --> C[portfolio-engine]
    B --> D[market-data-service]
    B --> E[sentiment-pipeline]
    B --> F[nli-interface]
    B --> G[analysis-agent]
    B --> H[notification-service]
    D -->|WebSocket binary| I["Dhan Live Feed\nwss://api-feed.dhan.co"]
    D -->|REST fallback| J[Dhan REST API]
    E --> K[(PostgreSQL + PGVector)]
    C --> K
    E -->|RAG| L[OpenAI Embeddings]
    F -->|Tool Calling| M[OpenAI GPT-4o]
    G -->|Multi-step Advisor| M

Loading

---

File Structure

PortfolioManager/
├── pom.xml                        ← Root Maven multi-module POM
├── .env.example                   ← Copy → .env and fill secrets
├── docker-compose.yml             ← One-command stack launch
├── scripts/init-pgvector.sql      ← Enables pgvector, hstore, uuid-ossp
│
├── common/                        ← Shared DTOs, enums, exceptions
├── portfolio-engine/              ← TIER 1: XIRR, PortfolioService, Flyway
├── market-data-service/           ← TIER 1: Dhan WebSocket + REST, Caffeine
├── sentiment-pipeline/            ← TIER 2: PGVector RAG + news scraper
├── nli-interface/                 ← TIER 2: Spring AI Tool Calling
├── analysis-agent/                ← TIER 3: Multi-step ChatClient advisor
├── notification-service/          ← Daily 9AM email: Spring Mail + Thymeleaf
├── api-gateway/                   ← Spring Boot app: security, controllers
│   └── src/main/resources/
│       └── application.yml        ← All configuration (DB, AI, Dhan, JWT)
│
└── frontend/                      ← Angular 19
    ├── src/app/
    │   ├── store/portfolio/       ← NgRx: portfolio feature slice
    │   ├── store/agent/           ← NgRx: agent/chat feature slice
    │   ├── core/services/         ← PortfolioService (typed HTTP client)
    │   ├── components/
    │   │   ├── net-worth/         ← D3.js treemap (Stocks > MF > NPS > EPF > PPF > FD)
    │   │   ├── live-console/      ← Analyze mode + NLI chat mode
    │   │   └── daily-recs/        ← AI recommendation cards
    │   └── pages/dashboard/       ← Main dashboard page
    ├── tailwind.config.js
    ├── nginx.conf
    └── Dockerfile

---

Getting Started

1. Fill in secrets

cp .env.example .env
# Edit .env with your: Dhan credentials, OpenAI API key, Gmail app password, JWT secret

2. Start the database

docker compose up postgres -d
# Wait for: "wealthengine-postgres is healthy"

3. Run the backend

cd /Users/adithya/Projects/PortfolioManager
mvn install -DskipTests
mvn spring-boot:run -pl api-gateway
# Backend: http://localhost:8080
# Swagger UI: http://localhost:8080/swagger-ui.html

4. Run the frontend

cd frontend
npm install
npm start
# Frontend: http://localhost:4200

5. (Optional) Full Docker stack

docker compose up --build
# Frontend: http://localhost:4200  |  Backend: http://localhost:8080

6. Default login

• Username: admin
• Password: changeme

---

Key Technical Decisions

Decision	Choice	Rationale
Market data	Dhan WebSocket + REST fallback	Live ticks via binary WS; REST for snapshots when WS down
LLM	OpenAI gpt-4o	Best tool-calling quality; swap to Gemini via 1 config line
Embeddings	text-embedding-3-small (1536d)	Good quality, cost-effective for news RAG
Cache	Caffeine (500 entries, 5s TTL)	Sub-ms latency for repeated tick lookups
Circuit breaker	Resilience4j	Auto-recovers from Dhan API outages
Migrations	Flyway	Schema as code, reproducible DB state

---

Verification Tests

# Run XIRR unit tests (9 test cases)
mvn test -pl portfolio-engine -Dtest=XirrCalculatorTest

# Full build (skip tests)
mvn clean install -DskipTests

# Health check
curl http://localhost:8080/actuator/health

# Login and get JWT
curl -X POST http://localhost:8080/api/v1/auth/login \
  -H 'Content-Type: application/json' \
  -d '{"username":"admin","password":"changeme"}'

# Portfolio summary (with token)
curl -H 'Authorization: Bearer <token>' http://localhost:8080/api/v1/portfolio/summary

# Analyze a stock
curl -X POST -H 'Authorization: Bearer <token>' \
  http://localhost:8080/api/v1/analyze/INFY

---

What Requires API Keys Before Working

Feature	Key Required
Live market ticks	DHAN_ACCESS_TOKEN + DHAN_CLIENT_ID
News sentiment (RAG)	OPENAI_API_KEY
NLI chat queries	OPENAI_API_KEY
AI stock analysis	OPENAI_API_KEY
Daily email reports	Gmail SMTP credentials

Tip

Everything except AI features works without any API keys. Add holdings to the DB manually or via Swagger to test the portfolio engine and XIRR calculations immediately.