← Back to case studies
Confidential Client
1B+ Record AI-Assisted Metadata Enrichment Platform
Records unified
1B+
Query interface
SQL only
→
natural language
Data quality
manual
→
automated expectations
A global media and entertainment company needed a unified, AI-enriched metadata platform to consolidate data from multiple major streaming and social platforms.
What changed
Before
Sources
Per-platform feeds
inconsistent
Access
SQL only
analysts blocked
After
Ingest
Delta Live Tables
quality expectations
Layers
Bronze
Silver
Gold
1B+ records
Query
AI-assisted
natural language
Consume
Domo dashboards
global
Streaming platform feeds consolidated into a medallion lakehouse, with the gold layer queryable in plain English.
What we built
The client needed to unify metadata from Apple EPF, Spotify, YouTube, and TikTok into a single enriched source of truth. Fragmented data across platforms was blocking marketing, rights management, and analytics teams from accessing reliable, timely information.
Key Features
Medallion architecture on Databricks Delta Lake integrating Apple EPF, Spotify, YouTube, and TikTok.
Spark ETL pipelines processing 1B+ metadata records with ML embedding enrichment for semantic search.
Databricks AI/Genie integration for natural language data exploration across the gold layer.
Delta Live Tables for continuous, incremental data quality enforcement.
LLM-powered unstructured text to structured model conversion for enrichment tasks.
Advanced BI dashboards with filtering across markets, rights, and analytics dimensions.
Data governance via Databricks Unity Catalog with role-based access for marketing, rights, and analytics teams.
Results
1B+ record unified metadata source consolidating all major streaming platforms.
AI-assisted analytics enabled non-technical users to query the gold layer using natural language.
Real-time BI dashboards deployed for marketing, rights management, and analytics globally.
Data quality issues identified and resolved automatically via Delta Live Tables expectations.
Stack
Engagement: 4 years · Real-time & Industrial Data Platforms, Product Engineering, AI Infrastructure & MLOps