What problem does it solve? Creating realistic test and demo data for Databricks is time-consuming, and naive random data lacks the patterns needed for dashboards, root cause analysis, and ML training. This Skill generates coherent, business-story-driven synthetic datasets at scale directly into Unity Catalog. ## Core Features & Use Cases - Scalable Spark + Faker Generation: Uses Databricks Connect serverless with Pandas UDFs to generate thousands to millions of rows in parallel, writing to Parquet, Delta, JSON, or CSV. - Realistic Data Patterns: Enforces non-uniform distributions (log-normal amounts, weighted categories, 80/20 rules), row coherence, time-based patterns, and referential integrity via master-table-first FK joins. - Story-Driven Planning Workflow: Presents a plan with business story, table specs, and assumptions for user approval before generating code, including optional bad-data injection for data quality testing. - Use Case: You need a support-ticket demo dataset showing an outage that spiked tickets and caused $2.3M churn. The Skill plans the tables, generates 80K tickets with correlated resolution times and CSAT, and writes them to your catalog volume. ## Quick Start Ask the assistant to generate a synthetic e-commerce dataset with customers and orders into a specific Unity Catalog and schema using serverless Spark.
