Human review onM

Demo workspace — safe sample data and reviewer workflow. No real applicants.

Sample data

Back to datasets

Rental Assistance Intake - March Pilot

Normalize source headers into durable storage keys and assign fraud-signal semantics before running vector analysis.

Map columns to signals

Storage-safe keys live in the database; the semantic mapping lives here.

8/8 mapped

Applicant Name
column_1
Signal type for Applicant Name
Service Address
column_2
Signal type for Service Address
Submission IP
column_3
Signal type for Submission IP
Phone
column_4
Signal type for Phone
Monthly Income
column_5
Signal type for Monthly Income
Payment Method
column_6
Signal type for Payment Method
Document Type
column_7
Signal type for Document Type
Routing Number
column_8
Signal type for Routing Number

Batch metadata

12,840 application rows

Row JSON is stored with storage-safe column names like column_1, while column_schema keeps the original header and signal mapping.

Applicant Name: column_1
Service Address: column_2
Submission IP: column_3
Phone: column_4
Monthly Income: column_5
Payment Method: column_6
Document Type: column_7
Routing Number: column_8