Infinity Dev
Skip to main content
Back to Projects

Project Gallery

1 of 5
RAG Keto Assistant Main Interface

About This Project

An intelligent AI-powered nutrition assistant that helps users with personalized keto diet recommendations using advanced RAG (Retrieval-Augmented Generation) technology. The system combines a Next.js frontend with a FastAPI backend, utilizing ChromaDB for vector storage and Google Gemini for AI responses.

Key Features

  • Multi-language support (English/Spanish) with next-intl internationalization
  • Voice recording with Web Speech API and browser compatibility detection
  • Real-time text-to-speech synthesis with audio playback controls
  • Chat interface with message history and real-time status indicators
  • Responsive design with mobile-first approach and cross-browser support
  • Modern dark theme UI with Shadcn/ui components and Tailwind CSS v4
  • Real-time AI streaming responses with token-by-token delivery
  • Semantic document retrieval using ChromaDB vector embeddings
  • Background TTS generation with Deepgram API integration
  • Intelligent query classification routing (literature vs recipes)
  • Async document ingestion pipeline with PDF processing

Technical Achievements

  • Full-stack integration between Next.js frontend and FastAPI backend with error handling
  • Real-time communication with streaming responses and audio synthesis
  • Browser compatibility with fallbacks for restrictive environments
  • Multi-language support with next-intl for UI and backend AI responses
  • Cross-browser testing with Playwright automation and visual regression
  • Production deployment with optimized builds and environment configurations
  • Accessible design with semantic HTML and ARIA implementations
  • Custom 404 error pages and responsive design patterns
  • Caching strategies and performance optimizations

Challenges & Solutions

  • Implementing real-time streaming responses while maintaining context
  • Building cross-browser voice recording with Web Speech API fallbacks
  • Optimizing vector database queries for sub-200ms response times
  • Creating seamless multilingual experience for both UI and AI responses

Architecture Overview

Microservices architecture with FastAPI backend and Next.js frontend, connected via RESTful APIs and Server-Sent Events for real-time streaming

Design Patterns

RAG (Retrieval-Augmented Generation)Repository PatternAsync/Await PatternEvent-Driven Architecture

Scalability

Server-Sent Events for real-time streamingVector database optimizationCDN integration for static assetsAsync processing pipelines

Links

Live Demo