Reverie

AI-powered Dream Analysis Platform for Personalized Insights

Reverie is an AI powered dream journaling platform for reflecting on emotions and uncovering patterns over time.

My team and I built it as a calm, intentional space for self reflection. Rather than offering instant interpretations or clinical conclusions, it encourages users to slow down, record their dreams, and engage thoughtfully with AI assisted insights.

Approach

Users write and revisit entries while exploring recurring emotional themes through structured analysis. The system focuses on pattern discovery and reflective summaries rather than prescriptive meaning, and the interface is deliberately minimal, encouraging consistent reflection rather than constant engagement.

Architecture

A full stack web application built around privacy and reliability. Dream entries are encrypted before storage, and the system stays stable even when AI services are unavailable.

Reverie's technology structure, from the Next.js frontend through the Python backend and AI layer to the PostgreSQL database
Technology structure

Stack

Area Tools
Frontend Next.js, TypeScript, Vercel
Backend Python, Flask, FastAPI, Cloud Run
AI layer TinyLLaMA, LoRA fine tuning, Hugging Face
Database PostgreSQL, Google Cloud SQL

Interface

The Reverie landing page
Landing page
The Reverie dashboard, showing recorded dream entries and emotional themes
Dashboard
The Reverie dream library, listing past entries for review
Dream library

Live

Reverie is live. You can join us by clicking the logo below.

Reverie logo, linking to the live site
Click our logo!

If you’re interested in our code, check out Reverie repository! 🚀