FoodAdvisor - AI-Powered Local Dietary Recommendation Platform

🎯 Project Overview

FoodAdvisor is an innovative dietary recommendation platform that connects users with personalized nutrition advice while leveraging real-time product data from local Dutch supermarkets. Our platform goes beyond generic meal suggestions by providing actionable, location-specific recommendations that users can immediately act upon.

Key Features

• Personalized User Dossiers: Each user maintains a comprehensive profile including dietary plans, requirements, preferences, and taste profiles
• AI-Powered Conversational Interface: Natural language interaction with intelligent agents that understand nutritional needs and local product availability
• Hyperlocal Recommendations: Direct integration with Dutch supermarket inventories (currently Jumbo, with plans for Albert Heijn, Action, Lidl and more)
• Actionable Guidance: Specific instructions like "Buy product X at Jumbo location Y, prepare using method Z"
• Comprehensive Planning: From quick meal suggestions to long-term nutrition strategies
• Budget & Nutrition Control: Full transparency and control over costs and nutritional intake

🛠️ Technical Architecture

Current Tech Stack

Data Pipeline

• Web Scraping: Custom scrapers collect product data from supermarket websites
• Storage: Amazon S3 for raw HTML storage
• ETL Processing: Python-based ETL pipeline (in progress) transforming HTML to structured data
• Database: PostgreSQL on AWS RDS with normalized schema for products, prices, and nutrition data

Infrastructure

• Cloud Provider: Amazon AWS
• Network: VPC configuration for secure internal communication
• Frontend: Angular framework (basic structure implemented)

Database Schema

• stores: Supermarket locations and metadata
• products: Product catalog with flexible unit system
• prices: Historical price tracking with promotions
• nutrition: Standardized nutritional information per 100g/100ml (Look backend/database/diagram.png for database outline)

System Architecture (Current)

[Web Scrapers] → [S3 Raw Storage] → [ETL Pipeline] → [PostgreSQL RDS]
                                                           ↓
                                                    [Angular Frontend]

📊 Current Status

✅ Completed

• Database schema design and implementation
• Web scraping infrastructure for Jumbo supermarket
• S3 storage pipeline for raw HTML data
• Basic Angular frontend structure
• Global Logger module
• Product data model with flexible unit system (weight/volume/piece/package)

🚧 In Progress

• ETL pipeline for HTML parsing and data transformation
• Frontend refinement

📋 Ready for Development

• RAG (Retrieval-Augmented Generation) pipeline
• AI agent architecture
• User authentication and profile management
• API Gateway and microservices migration to Amazon EC2.

🚀 Development Roadmap

Phase 1: Infrastructure Migration & Scalability

Goal: Transition from development VPC to production-ready architecture

1. EC2 Deployment

   • Migrate application services to EC2 instances
   • Implement auto-scaling groups for demand management
   • Configure load balancers for high availability

2. API Gateway Implementation

   • Design RESTful API endpoints for all services
   • Implement rate limiting and authentication
   • Create API documentation with OpenAPI/Swagger

3. Microservices Architecture

   • Separate concerns into distinct services:
     • Product Data Service
     • User Management Service
     • Recommendation Engine Service
     • Agent Communication Service

Phase 2: RAG Pipeline & AI Agent Development

Goal: Build intelligent recommendation system with conversational capabilities

1. RAG Infrastructure Setup

   • SQL query generation framework for PostgreSQL
   • Natural language to SQL translation layer
   • Query optimization for complex nutritional/price filtering
   • Result ranking and relevance scoring algorithms

2. Multi-Agent System Architecture

   Agent 1: Conversational Interface Agent

   • Primary user interaction point
   • Natural language understanding for dietary preferences
   • Query routing to specialized agents
   • Response synthesis and presentation

   Agent 2: User Profile Management Agent

   • Dossier creation and maintenance
   • Preference learning and adaptation
   • Historical tracking of user choices
   • Privacy-compliant data handling

   Agent 3: Nutrition & Diet Expert Agent (Future)

   • Deep knowledge of nutritional science
   • Dietary plan creation and optimization
   • Health goal alignment
   • Medical restriction compliance

3. Model Context Protocol (MCP) Implementation

   • Design inter-agent communication protocols
   • Implement shared context management
   • Create agent orchestration layer
   • Build fallback and error handling mechanisms

4. LLM Selection & Integration

   • Evaluate options: OpenAI GPT-4, Anthropic Claude, open-source alternatives
   • Implement prompt engineering framework
   • Create evaluation metrics for recommendation quality
   • Set up A/B testing infrastructure

Phase 3: User Management & Authentication

Goal: Secure, scalable user system with personalized experiences

1. Authentication System

   • Implement JWT-based authentication
   • OAuth2 integration for social login
   • Multi-factor authentication options
   • Session management

2. User Profile System

   • Comprehensive preference storage
   • Dietary restriction management
   • Goal tracking (weight, nutrition, health)
   • Shopping history and patterns

3. Privacy & Compliance

   • GDPR compliance implementation
   • Data encryption at rest and in transit
   • User data export capabilities
   • Right to deletion implementation

Phase 4: Advanced Features & Optimization

Goal: Enhanced user experience and system intelligence

1. Recommendation Engine Enhancements

   • Collaborative filtering for meal suggestions
   • Seasonal and weather-based adaptations
   • Cultural cuisine preferences
   • Family meal planning capabilities

2. Budget Optimization Module

   • Price tracking and alerting
   • Alternative product suggestions
   • Bulk buying recommendations
   • Store promotion integration

3. Nutritional Knowledge Base

   • Integration with scientific nutrition databases
   • Allergen and interaction warnings
   • Micronutrient tracking
   • Health condition specific recommendations

4. Multi-Store Integration

   • Albert Heijn scraper development
   • Price comparison features
   • Optimal shopping route planning
   • Inventory availability checking

5. Price History Analytics

   • Historical price trend analysis
   • Seasonal pricing pattern detection
   • Promotion frequency insights
   • Price prediction models
   • Best time to buy recommendations
   • Inflation impact tracking

Phase 5: Scale & Expansion

Goal: Platform growth and market expansion

1. Performance Optimization

   • Caching strategies implementation
   • Database query optimization
   • CDN integration for static assets
   • Real-time data synchronization

2. Analytics & Insights

   • User behavior analytics
   • Recommendation effectiveness metrics
   • A/B testing framework
   • Business intelligence dashboards

3. Mobile Applications

   • iOS and Android native apps
   • Offline capability for shopping lists
   • Barcode scanning integration
   • Push notifications for deals

4. Geographic Expansion

   • Additional Dutch supermarket chains
   • Belgian market entry preparation
   • Multi-language support
   • Local dietary preference adaptation

🏗️ Architecture Decisions to Make

1. LLM Provider Selection

   • Cost vs. performance analysis
   • Privacy and data residency requirements
   • Fine-tuning capabilities assessment

2. Agent Framework Selection

   • LangChain vs. custom implementation
   • AutoGPT/AutoGen consideration
   • Workflow orchestration tools

---

Project Status: 🟡 Active Development

Last Updated: December 2024