← Back to Sprints

Keepr

Live Project
Laravel 12Next.js 15Gemini AITanStack QueryAxiosAWS:ECS/Fargate

A sophisticated AI-powered document vault implementing RAG (Retrieval-Augmented Generation) and vector embeddings to provide semantic search and intelligent insights for personal notes.

Keepr

The Challenge

Standard note-taking applications act as static repositories, requiring manual organization and lacking the ability to provide deep context or cross-document intelligence.

The Solution

Architected a RAG pipeline using Google Gemini AI and vector embeddings to transform raw notes into a searchable knowledge base. The system utilizes a headless Laravel 12 backend for AI processing and TanStack Query on the frontend to manage complex server-state synchronization. Deployment leverages a high-availability AWS ECS sidecar architecture.

Project Access

Open Live ApplicationFrontend SourceBackend Source

Key Features

  • RAG (Retrieval-Augmented Generation) for context-aware document insights and summaries.
  • Semantic Search capabilities powered by vector embeddings for meaning-based retrieval.
  • Multi-container sidecar pattern (Nginx + PHP-FPM) on AWS ECS Fargate.
  • TanStack Query for robust client-side state management and background synchronization.
  • Secure PostgreSQL schema with JSONB support for hybrid structured/unstructured data.
  • AWS ALB with Host Header routing and SSL termination for secure cross-origin communication.