🏆 1st Place Winner – Databricks Global Hackathon
This project is an end-to-end Data & AI solution built entirely on Databricks Free Edition that automates YouTube content creation, publishing, discovery, and analytics using modern AI, LLMs, and Lakehouse architecture.
The solution was designed for a technical content creator (DataTuber) use case and demonstrates how multiple Databricks services can be combined into a single production-grade workflow.
- Upload video via a Databricks Streamlit Web App
- Convert video → audio (MoviePy)
- Generate transcript with timestamps (OpenAI Whisper – offline)
- Auto-generate:
- YouTube Title
- Description with chapters
- SEO Tags
- Thumbnail image (OpenAI Image Model)
- Automatically upload video to YouTube
- Send email notification to creator after upload
- Transcript is:
- Chunked with timestamps
- Stored in Delta tables
- Embedded using Databricks Foundation Models
- Embeddings stored in Databricks Vector Search
- Users can ask natural language questions
- System returns:
- Summarized answer
- Direct YouTube link with exact timestamp
- Implements Retrieval Augmented Generation (RAG) architecture
- Fetch YouTube metrics using YouTube APIs:
- Views, likes, comments
- Perform sentiment analysis on comments using Databricks SQL AI functions
- Store structured data in Delta tables
- Interactive:
- Databricks Dashboards
- AI/BI Genie Workspace for natural language analytics
- Business users can query performance without writing SQL
Personas Supported
- Media Creator – Uploads videos and receives automated publishing
- Knowledge Explorer – Searches content via AI-powered Q&A
- Business Owner – Analyzes channel performance via dashboards
Core Technologies
- Databricks Free Edition
- Delta Lake
- Databricks Jobs
- Databricks Vector Search
- Databricks Foundation Models
- Streamlit (Databricks App)
- OpenAI Whisper (offline)
- OpenAI Image Generation
- YouTube Data APIs
- Retrieval Augmented Generation (RAG)