Alternative Data Intelligence Dashboard

An AI-powered analytics platform for discovering correlations between alternative data signals (job postings, social media sentiment, employee growth) and stock performance. Features natural language queries, dual visualization modes, and an intelligent discovery engine for identifying predictive patterns.

🚀 Quick Links

Deploy & Auth Setup → Complete Setup Guide - Production deployment
Demo Prep → docs/DEMO_CHEAT_SHEET.md - Interview talking points
Architecture → docs/ARCHITECTURE.md - Technical deep dive

🌟 Key Features

📊 Dual Visualization Modes

Correlation View: Interactive scatter plots showing metric relationships with Pearson correlation coefficients
Trend View: Time-series charts with dual Y-axes for temporal pattern analysis
Toggle between visualization modes with one click
Color-coded multi-stock comparisons (up to 3 stocks)
Interactive tooltips with detailed data points
12 tickers: AAPL, AMZN, DELL, GOOGL, JNJ, META, MSFT, NKE, NVDA, TSLA, UBER, V
11 alternative data metrics with intelligent display names

🤖 Natural Language Query Interface

Plain English queries: "Show correlation between job postings and price for AAPL"
Multi-stock comparison: "Compare TSLA vs NVDA Reddit sentiment" (up to 3 stocks)
AI-powered parsing: Claude Haiku 4.5 extracts structured parameters with 90% cost savings vs Sonnet
Interactive query editing: Click pill-shaped tags to modify tickers and metrics
Pending changes system: Preview and update queries before execution
Smart validation: Multi-layer error handling with actionable suggestions
Real-time feedback: Loading states, confidence scoring, and parsed parameter display

🔍 Auto-Discovery Engine

One-click discovery: Automatically test all metrics against stock price
Correlation ranking: Results sorted by predictive strength
Interactive results: Click any discovery result to visualize the correlation
Expandable view: Show top 5 or all correlations with strength indicators
Parallel processing: Fast correlation calculations across all metrics

🎨 Interactive Query Builder

Visual query editing: Pill-based UI for modifying tickers and metrics
Add/remove tickers: Dynamic ticker management with visual feedback
Metric obfuscation: User-friendly display names hide internal data sources
Pending changes preview: See changes before applying them
Smart suggestions: Context-aware recommendations for query improvements

🔐 Secure Authentication & Authorization

Email/password authentication: Powered by Supabase Auth
Admin approval workflow: New users require approval before accessing data
Real-time status updates: Instant UI updates when approval status changes
Email notifications: Automated admin alerts and user welcome emails via edge functions
Row Level Security: Database-level access controls

🚀 Quick Start

Prerequisites

bun >= 1.0.0  # Recommended: Fast JavaScript runtime
# OR node >= 18.0.0 (if you prefer npm)

Installation

bun install  # Recommended (faster)
# OR: npm install

Environment Variables

Create a .env.local file with the following variables:

# AI Features
ANTHROPIC_API_KEY=your_anthropic_key

# Supabase
NEXT_PUBLIC_SUPABASE_URL=https://your-project.supabase.co
NEXT_PUBLIC_SUPABASE_ANON_KEY=your_anon_key
SUPABASE_SERVICE_ROLE_KEY=your_service_role_key

# Approval Workflow
APPROVAL_SECRET=generate_random_string
NEXT_PUBLIC_SITE_URL=http://localhost:3000
[email protected]

See CLAUDE.md for detailed environment variable documentation.

Run Development Server

bun run dev  # Recommended (faster startup)
# OR: npm run dev

Open http://localhost:3000 in your browser.

Build for Production

bun run build  # Recommended (faster builds)
bun run start
# OR: npm run build && npm start

📖 Documentation

Comprehensive documentation is available in the docs/ directory:

For Users

FEATURES.md - High-level feature overview and use cases
USAGE_GUIDE.md - Complete user guide with query examples

For Developers

ARCHITECTURE.md - Technical architecture and data flow
NATURAL_LANGUAGE_QUERIES.md - API documentation and implementation details
IMPLEMENTATION_SUMMARY.md - Complete implementation summary
BUN_OPTIMIZATIONS.md - Bun runtime guide and performance benchmarks

For Deployment

DEPLOYMENT_CHECKLIST.md - Pre-deployment verification steps

For Presentations

DEMO_CHEAT_SHEET.md - Quick reference for product demonstrations

🧪 Testing

Run the API test suite:

# Terminal 1: Start dev server
bun run dev

# Terminal 2: Run tests
bun run test:api
# OR: node test-api.js

💡 Example Queries

Try these natural language queries:

Single Stock Analysis

Show correlation between job postings and price for AAPL
Does Twitter engagement predict NVDA stock movement?
Show me employment signals vs price for META since 2024

Multi-Stock Comparison (NEW)

Compare TSLA vs NVDA Reddit sentiment
Compare AAPL, MSFT, and GOOGL job postings vs price
How do TSLA and NVDA differ on Twitter mentions?

🛠️ Tech Stack

Runtime: Bun 1.x (recommended) or Node.js 18+
Framework: Next.js 16.0.1 with App Router
Language: TypeScript 5
UI: React 19.2.0 + Tailwind CSS 4
Charts: Chart.js 4.5.1 + react-chartjs-2
AI: Anthropic Claude Haiku 4.5 (@anthropic-ai/sdk 0.68.0)
Database: Supabase (PostgreSQL)
Deployment: Vercel-ready

📁 Project Structure

alt-data-dashboard/
├── src/
│   ├── app/
│   │   ├── api/
│   │   │   ├── parse-query/route.ts    # Natural language parser
│   │   │   ├── correlation/route.ts     # Single-stock correlation
│   │   │   └── compare/route.ts         # Multi-stock comparison (NEW)
│   │   ├── components/
│   │   │   └── CorrelationChart.tsx     # Main chart component
│   │   ├── layout.tsx
│   │   └── page.tsx
│   ├── lib/
│   │   └── supabase.ts                  # Supabase client
│   └── types/
│       └── query.ts                      # TypeScript types
├── scripts/
│   └── import_data.py                    # Data import script
└── Documentation files (*.md)

🎯 Key Implementation Details

Natural Language Processing

User enters query in search box
Query sent to /api/parse-query
Claude AI extracts: ticker(s), metricX, metricY, dates
Validation against whitelists (max 3 tickers enforced)
Structured parameters sent to /api/correlation (single) or /api/compare (multi)
Chart updates with correlation data (color-coded for multi-stock)

Error Handling

Multi-layer validation (client, API, AI response)
User-friendly error messages
Helpful suggestions for fixing queries
Graceful fallback to default view

Data Pipeline

Python script imports Excel files to Supabase
~12,080 records across 12 tickers
Date range: 2024-present
Handles NaN values and duplicates

🚢 Deployment

Vercel (Recommended)

Push code to GitHub
Import repository in Vercel
Add environment variables in Vercel dashboard
Deploy

Environment Variables in Production

Set these in your deployment platform:

ANTHROPIC_API_KEY
NEXT_PUBLIC_SUPABASE_URL
SUPABASE_SERVICE_ROLE_KEY

📊 Available Data

Tickers (12)

AAPL, AMZN, DELL, GOOGL, JNJ, META, MSFT, NKE, NVDA, TSLA, UBER, V

Metrics (11)

Metrics are displayed with user-friendly names that hide internal data sources:

Stock Price - Daily closing price
Job Postings - Aggregated job posting counts
Community Activity Index - Reddit post mentions (reddit_mentions)
Social Velocity - Twitter/X mentions (twitter_mentions)
Sentiment Alpha - Reddit sentiment score (reddit_sentiment)
Social Reach Metric - Twitter follower count (twitter_followers)
Workforce Index - LinkedIn employee count (employees_linkedin)
Hiring Momentum Score - AI employment signal (ai_score_employment)
Composite Signal - Overall AI score (ai_score_overall)
Investor Sentiment Index - StockTwits sentiment (stocktwits_sentiment)
News Mentions - News article mentions

Data Volume

~12,080 total records across all tickers
~1,000 data points per ticker
Date range: 2024-present
Daily granularity with null value handling

🔐 Security

API keys never exposed to client
Server-side validation of all inputs
Supabase Row Level Security (RLS)
TypeScript for type safety
Input sanitization

🎥 Feature Demonstration

Quick demonstration flow (< 3 minutes):

Natural Language Query: Execute a plain English query
Discovery Engine: Click "Discover" to find top correlations
Interactive Editing: Click pills to modify query parameters
Dual Visualizations: Toggle between Correlation and Trend views
Multi-Stock Comparison: Compare up to 3 stocks simultaneously
Error Handling: See intelligent error messages and suggestions

See DEMO_CHEAT_SHEET.md for detailed presentation guide.

📈 Visualization Modes

Correlation View

Scatter plot showing relationship between two metrics
Each point represents one day's data
Visual correlation strength with color coding
Available for all metric combinations

Trend View (NEW)

Time-series line charts with dual Y-axes
Left axis: Stock price (always)
Right axis: Alternative data metric
Track both metrics over time simultaneously
Requirement: One metric must be price
Supports both single-stock and multi-stock comparisons
Color-coded lines for easy distinction

Multi-Stock Trend View:

Price lines: Blue, Red, Purple (one per stock)
Metric lines: Green, Yellow, Teal (one per stock)
Legend clearly identifies each line
All stocks shown on same timeline

🚀 Potential Enhancements

The following features could extend the platform's capabilities:

Query History & Favorites: Save and recall frequently used queries
Voice Input: Speech-to-text for hands-free query input
Advanced Statistics: R², p-values, confidence intervals, and trend lines
Export Functionality: Download results as CSV, PNG, or PDF reports
Query Caching: Redis-based caching for improved performance
Real-time Data: WebSocket integration for live data updates
Extended Comparison: Support for more than 3 stocks with tabbed interface
Custom Alerts: Notification system for correlation threshold breaches
API Access: RESTful API for programmatic access to correlation engine

📝 License

MIT License

🙏 Acknowledgments

This project leverages best-in-class technologies:

Next.js & React 19 - Modern web framework with experimental compiler optimizations
Anthropic Claude Haiku 4.5 - Cost-effective AI with reliable JSON output
Supabase - PostgreSQL database with real-time capabilities and authentication
Chart.js - Flexible visualization library for correlation and time-series charts
Tailwind CSS 4 - Utility-first CSS framework
Bun Runtime - High-performance JavaScript runtime (3-4x faster than Node.js)

🎯 Technical Highlights

O(n) Correlation Algorithm: Optimized single-pass Pearson coefficient calculation
React 19 Compiler: Experimental automatic memoization and optimization
Three-Tier Supabase Architecture: Separate clients for service role, SSR, and browser contexts
Abort Controller Pattern: Prevents race conditions in concurrent API requests
Real-time Subscriptions: Instant UI updates via Supabase real-time
Multi-Layer Validation: Client, API, AI, and database validation for robustness
Cost Optimization: 90% savings using Haiku 4.5 vs Sonnet with comparable performance

Tech Stack: Next.js 16 • React 19 • TypeScript 5 • Claude Haiku 4.5 • Supabase • Bun Features: AI-Powered NLP • Dual Visualizations • Auto-Discovery • Interactive Editing Status: ✅ Production Ready

Name		Name	Last commit message	Last commit date
Latest commit History 29 Commits
bruno		bruno
docs		docs
public		public
scripts		scripts
src		src
.gitignore		.gitignore
README.md		README.md
bun.lock		bun.lock
eslint.config.mjs		eslint.config.mjs
middleware.ts		middleware.ts
next.config.ts		next.config.ts
package.json		package.json
postcss.config.mjs		postcss.config.mjs
test-api.js		test-api.js
tsconfig.json		tsconfig.json

drsullivan13/alt-data-dashboard

Folders and files

Latest commit

History

Repository files navigation