Solutions · Databricks

A lakehouse with no raw identifiers.

The native Databricks connector tokenizes at the source or the ingestion job: notebooks, features, and models all operate on tokens, with detokenization restricted to governed service identities.

Buyer pains

Why this buyer hurts.

Lakehouse of PII

Raw drops land identifiers in the cheapest, most replicated storage you own.

Notebook sprawl

Interactive access means PII in cells, MLflow artifacts, and ad-hoc exports.

Feature leakage

Identifiers slip into feature stores and ride into models you cannot un-train.

How cloakit fixes it

Token-first ingestion, natively.

cloakit tokenizes at the source or the ingestion job: Databricks jobs, notebooks, and models all operate on tokens, with detokenization restricted to governed service identities.

Native Databricks connector: tokenized landing zones, governed service-identity access, and identifier-free feature stores.

How it runs

The workflow, end to end.

Land tokens in bronze

Ingestion jobs resolve identifiers to tokens before the first write — raw values never touch storage.

Build features safely

Silver and gold layers, notebooks, and feature stores compose on token keys with full lineage.

Train without identity

Models learn from tokenized features; any re-identification is a governed vault call, not a dataframe read.

Questions

Asked by teams like yours.

Do notebooks change?

No — analysts query the same tables and columns; values are simply tokens outside governed resolution.

What about MLflow artifacts?

Artifacts inherit the tokenized lineage, so experiment tracking stays shareable without scrubbing.

Who can detokenize?

Only governed service identities under policy — never interactive users by default.

See it on your own workflow.

Bring one sensitive flow. We will map it to a vault schema live in 30 minutes.

Get a Demo See the API pattern

Lakehouse path

Tokenize at the first write.

Raw drops are tokenized by the ingestion job before bronze. Notebooks, features, and models operate on lineage-carrying tokens throughout.

%%{init: {'flowchart': {'curve': 'linear'}, 'theme': 'base', 'themeVariables': {'primaryColor': '#ffffff', 'primaryBorderColor': '#2457C5', 'primaryTextColor': '#101828', 'lineColor': '#2457C5', 'fontFamily': 'IBM Plex Sans, sans-serif', 'fontSize': '14px'}}}%% flowchart LR S["Raw drops"] --> T["Ingest job
tokenize first"] T --> B["Bronze / Silver
token lineage"] B --> F["Features + models
no identifiers"] F --> Q{"Resolve?"} Q -- service identity + policy --> V["Governed resolve"] Q -- interactive user --> X["Denied by default"]

Native connector

Databricks

The native connector tokenizes before bronze — every layer above inherits tokens, lineage, and policy.

Also connects

Snowflake

BigQuery

ClickHouse

DuckDB

Trino

Third-party names and marks are property of their respective owners, listed for compatibility.