masking-pii-datalisted

# Masking PII Data ## When to use - A pipeline or table contains personal/sensitive data (names, emails, SSNs, payment, health). - Choosing how to de-identify data for analytics or lower environments. - Restricting who can see raw sensitive columns. - Handling deletion / right-to-be-forgotten requests. - Do NOT use for general access control unrelated to sensitive data. ## Choose the right technique | Technique | Reversible | Keeps analytics utility | Use for | | ------------------- | --------------- | ----------------------- | ------------------------------------- | | Masking / redaction | No | Low | Display, lower envs (`j***@x.com`) | | Hashing (salted) | No | Join/match only | Pseudonymous keys, dedup | | Tokenization | Yes (via vault) | Referential joins | Reversible pseudonymization | | Encryption | Yes (with key) | None until decrypt | At-rest protection, restricted fields | ## Workflow ``` - [ ] Classify columns: what is PII/sensitive and its risk level - [ ] Pick technique per column by whether you need reversibility/joins - [ ] Apply as early as possible (mask on ingest for lower environments) - [ ] Enforce column-level access / dynamic masking for raw data - [ ] Support deletion: know every location a subject's data lives ``` 1. **Classify first.** You can't protect what you haven't identified; tag colu