Dream Data Lakehouse - Sachi Parikh

Demo Video

System Architecture

Repository Structure

📁 dream-data-lakehouse-bedrock/
├── 📁 frontend/ → React/TypeScript web app
├── 📁 backend/ → FastAPI Python backend
├── 📁 infrastructure/ → AWS CDK/Terraform configs
├── 📁 notebooks/ → Jupyter notebooks for analysis
├── 📁 scripts/ → ETL and data processing scripts
└── 📄 README.md

Key Features

Bedrock Integration: Leverages Claude for synchronous text generation (movie scripts, analysis) and Luma Ray 2 for asynchronous video generation
S3 Lakehouse: Raw event logs stored in partitioned JSON format with append-only architecture
ETL Pipeline: AWS Glue transforms raw data into curated Parquet format for efficient querying
Analytics: Amazon Athena enables SQL queries on dream patterns, moods, motifs, and outcomes
Async Video Processing: Poll-based status checking for Luma Ray 2 video generation with S3 output delivery
FastAPI Backend: RESTful API for dream journal entries, render job management, and status polling
Theater UI: Immersive curtain-loading experience with autoplay video and calendar-based dream log