PrismSekData Platform
PrismSek for Databricks
Lakehouse and ML pipeline coverage where data science meets regulated data.
Why it matters
Databricks is where raw data becomes features, models, and AI products. PrismSek classifies Delta tables and volumes, audits training corpora, and keeps regulated data out of notebooks and fine-tunes.
Visibility and control
What we see
- Unity Catalog tables, volumes, and schemas
- Delta Lake content at column level
- Notebook outputs and workspace files
- ML pipelines, feature tables, and model artifacts
What we act on
- Unity Catalog tag sync from PrismSek classifications
- Training-corpus audits before fine-tuning jobs
- Pipeline gates that exclude or redact regulated records
- Lineage from raw tables to model artifacts
Risks this closes
- Regulated columns flowing into feature stores unlabeled
- Notebooks printing sensitive samples into outputs
- Training sets assembled from ungoverned raw zones
- Workspace files accumulating exported extracts
Connector details
Connector type
Native connector via workspace service principal, agentless
Authentication
Service principal with Unity Catalog metadata + sampled-read
Scan scope
Metastore-wide or per catalog/schema, sampling configurable
Products active in this environment
Connect Databricks in a guided session.
Most environments show first findings within minutes of authenticating the connector.