YouTube Wrapped

A personal "Spotify Wrapped" for YouTube watch history. The project turns a Google Takeout export into a polished year-in-review dashboard with Databricks, Neon Postgres, FastAPI, and Next.js.

Live demo | API docs

[!NOTE]

Why I Built This

YouTube watch history is full of patterns, but the raw Takeout export is not built for exploration. This project cleans and enriches that data, then presents it as a shareable analytics experience: top artists, music share, genre split, binge sessions, listening rhythm, and loyalty insights.

Preview

Personal insights	Listening rhythm

Genres and loyalty	Databricks pipeline

Highlights

• End-to-end data product from Google Takeout JSON to deployed web dashboard.
• Medallion lakehouse pipeline with Bronze, Silver, Enrichment, and Gold notebooks in Databricks.
• Typed FastAPI service backed by Neon Postgres fact tables.
• Next.js dashboard with animated cards, responsive layouts, and cached API calls.
• Music enrichment through YouTube metadata and MusicBrainz-style artist/genre normalization.
• Deployment split across Vercel for the frontend, Render for the API, and Neon for the serving database.

Architecture

flowchart TD
    subgraph Data_Source ["📤 Data Ingestion"]
        Takeout["Google Takeout (watch-history JSON)"]
    end

    subgraph Databricks_Medallion ["💎 Databricks Medallion Lakehouse"]
        Bronze[("🥉 Bronze Layer\nRaw Data Landing")]
        Silver[("🥈 Silver Layer\nCleaned, Typed, Deduplicated")]
        Enrichment["⚡ Enrichment Engine\nYouTube Metadata & Music Mapping"]
        Gold[("🥇 Gold Layer\nAnalytics Fact Tables")]
    end

    subgraph Serving_Layer ["🚀 Serving & API Layer"]
        Neon[("🐘 Neon PostgreSQL\nServing Database")]
        FastAPI["⚡ FastAPI Backend\n(Hosted on Render)"]
    end

    subgraph Presentation_Layer ["🖥️ Presentation Layer"]
        NextJS["🌐 Next.js Dashboard\n(Hosted on Vercel)"]
    end

    Takeout -->|"Ingest JSON"| Bronze
    Bronze -->|"Parse & Clean"| Silver
    Silver -->|"Add Metadata"| Enrichment
    Enrichment -->|"Aggregate Facts"| Gold
    Gold -->|"Load via Script"| Neon
    Neon -->|"SQL Query"| FastAPI
    FastAPI -->|"REST /api"| NextJS

    style Takeout fill:#ff0000,stroke:#333,stroke-width:2px,color:#fff
    style Bronze fill:#cd7f32,stroke:#333,stroke-width:2px,color:#fff
    style Silver fill:#c0c0c0,stroke:#333,stroke-width:2px,color:#000
    style Gold fill:#ffd700,stroke:#333,stroke-width:2px,color:#000
    style Neon fill:#336791,stroke:#333,stroke-width:2px,color:#fff
    style FastAPI fill:#009688,stroke:#333,stroke-width:2px,color:#fff
    style NextJS fill:#000000,stroke:#333,stroke-width:2px,color:#fff

Loading

Tech Stack

Layer	Tools
Data source	Google Takeout YouTube watch history
Lakehouse	Databricks Free Edition, Unity Catalog, Delta Lake
Transformation	PySpark, Databricks notebooks, medallion architecture
Enrichment	YouTube Data API, music metadata classification
Serving database	Neon Postgres
API	FastAPI, SQLAlchemy, Uvicorn
Frontend	Next.js 16, React 19, TypeScript, Tailwind CSS 4, Recharts, Framer Motion
Deployment	Vercel, Render, Neon

Project Structure

youtube-wrapped/
|-- api/                 # FastAPI service
|   |-- app/
|   |   |-- main.py      # App setup, CORS, router registration
|   |   |-- database.py  # Neon/Postgres connection
|   |   |-- models.py    # Pydantic response models
|   |   `-- routers/     # Analytics endpoints
|   `-- requirements.txt
|-- frontend/            # Next.js app
|   |-- src/app/         # App Router pages and global styles
|   |-- src/components/  # Dashboard cards and interactions
|   `-- src/lib/api.ts   # Typed API client
|-- notebooks/           # Databricks pipeline notebooks
|   |-- bronze/          # Raw ingestion
|   |-- silver/          # Clean, typed, deduplicated data
|   |-- enrichment/      # Metadata and music enrichment
|   `-- gold/            # Analytics fact tables
|-- scripts/
|   `-- load_to_neon.py  # Loads gold CSV exports into Neon
|-- docs/screenshots/    # README showcase images
`-- render.yaml          # Render API deployment config

Analytics Included

• Overview totals: total watches, music share, unique artists, days tracked.
• Main character artist: the artist that defined the listening year.
• Top artists, channels, and genres.
• Genre split across Desi, Western, and untagged music.
• Listening rhythm by hour and day.
• Binge sessions with video count and duration.
• Loyal artists ranked by listening span.
• Last pipeline run timestamp surfaced in the dashboard footer.

Run Locally

Backend

cd api
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
uvicorn app.main:app --reload

Create an API environment file before running against Neon:

NEON_CONNECTION_STRING=postgresql://user:password@host:port/database

The API runs at http://localhost:8000, with Swagger docs at http://localhost:8000/docs.

Frontend

cd frontend
npm install
npm run dev

Create frontend/.env.local if your API is not running at the default local URL:

NEXT_PUBLIC_API_URL=http://localhost:8000

The dashboard runs at http://localhost:3000.

Data Workflow

1. Export YouTube watch history from Google Takeout.
2. Run notebooks/bronze/02_bronze_ingest.ipynb to land the raw history in Databricks.
3. Run notebooks/silver/03_silver_clean.ipynb to parse timestamps, clean titles, type columns, and deduplicate rows.
4. Run notebooks/enrichment/04_enrich_youtube.ipynb to add video, channel, artist, and genre context.
5. Run notebooks/gold/05_gold_facts.ipynb to produce dashboard-ready fact tables.
6. Export fact_*.csv files into data/gold_exports/.
7. Load the serving tables into Neon:

python scripts/load_to_neon.py

Raw exports, CSVs, and secrets are intentionally ignored by Git so personal watch history is not committed.

API Surface

The FastAPI app exposes read-only analytics endpoints under /api, including:

/api/overview
/api/top-artists
/api/top-channels
/api/top-genres
/api/genre-split
/api/listening-by-hour
/api/listening-by-dayofweek
/api/timeline
/api/main-character
/api/binge-sessions
/api/night-owl-score
/api/loyal-artists
/api/last-pipeline-run

Deployment

• Frontend: deployed on Vercel.
• Backend: deployed on Render using render.yaml.
• Database: hosted on Neon Postgres.
• Pipeline: run in Databricks, then loaded into Neon with scripts/load_to_neon.py.

License

MIT License. See LICENSE.