Synthetic proof · Data QA sprint

Eight messy rows. Five actionable findings. One corrected artifact.

This small, reproducible example shows the output of a bounded CSV/JSON validation sprint for teams that need a clean handoff—not a vague “data looks fine” assurance.

Evidence boundary: this is self-directed synthetic data, not client work or a claim about production outcomes. No customer data, credentials, uploads, or external systems are used.

input rows

Order-like records created only for this sample.

ranked findings

Schema, duplicate, missing-field, date, and outlier checks.

clean output rows

Duplicate removed; invalid record quarantined.

Representative input

order_id	created_at	sku	quantity	status
ORD-1001	2026-07-20	SKU-A	2	shipped
ORD-1001	2026-07-20	SKU-A	2	duplicate
ORD-1003	2026-02-30	SKU-C	1	pending
ORD-1004	2026-07-21	(missing)	3	paid
ORD-1005	2026-07-21	SKU-B	999	paid

Ranked findings and disposition

Severity	Finding	Evidence	Disposition
High	Duplicate primary key	`ORD-1001` occurs twice	Keep first record; quarantine duplicate
High	Invalid calendar date	`2026-02-30`	Quarantine pending source correction
High	Required value missing	`sku` empty for ORD-1004	Flag for source enrichment
Medium	Quantity outlier	`999` exceeds sample rule of 100	Retain but require review
Medium	Unexpected enum value	`complete` outside approved status set	Map only after owner confirmation

Smallest sellable sprint

USD 150 for one CSV or JSON export, up to 10,000 records and one agreed schema.

Includes duplicate/missing/type/date/outlier checks, a ranked report, corrected artifact where deterministic, and one review pass. No production integration.

What the buyer receives

Input/profile summary
Evidence-linked findings
Corrected and quarantined records
Validation rules and limitations

Request a free fit check

Describe the format, approximate row count, and validation goal. Do not email customer data, credentials, or the dataset itself.