DATAMIMIC: Deterministic test data for regulated enterprises

Open-source synthetic data engine with AI agent integration, deterministic output, and enterprise governance for banking and healthcare.

18 cards · 3m · tap to begin

From · · 3m

DATAMIMIC: Deterministic test data for regulated enterprises

Open-source synthetic data engine with AI agent integration, deterministic output, and enterprise governance for banking and healthcare.

In brief

Open-source synthetic data engine with AI agent integration, deterministic output, and enterprise governance for banking and healthcare. Same seed plus same model always produces byte-identical output, making test data reproducible across runs and machines. Originally reported by GitHub.

Two editions, one determinism contract

Community Edition is MIT-licensed Python for synthetic data and pseudonymization. Enterprise Platform adds PII scanning, multi-system execution, governance, and audit trails.

Built for AI agents from day one

CLI exposes machine-readable capabilities, progressive reference queries, and a single scaffold transaction. Optional MCP adapter maps the same operations to tool calls.

Determinism means byte-identical output

Same engine version, same model, same seed produces identical output every run on every machine. Verified per-service on every CI run.

“Same engine version + same model + same seed = byte-identical output, every run, every machine.”

Domain services know real constraints

Healthcare generates age-appropriate conditions. Finance produces balance-consistent accounts. Faker gives you a 25-year-old with Alzheimer's.

Seeded pseudonymization for regression tests

Set rngSeed and the same source record maps to the same synthetic value every run. Drop the seed for privacy-maximized one-time delivery.

PII field detection is manual in CE

You identify and model PII fields in the XML pipeline. Enterprise Platform automates this with a probability-scored scanner and DataWorkbench.

Time-series generation with ISO 8601

Any generate block becomes a time-series loop with start, end, and interval attributes. Output is prefix-stable by construction.

Three reasons teams adopt CE standalone

Reproducible test data for CI/CD. Deterministic backend for AI agents with verifiable content hashes. Pseudonymization of staging exports without a scanner license.

Enterprise Platform adds operational scale

Rust fastpath, ML engine, keyset building from live schemas, and billion-record workloads. Three runtime profiles: Performance, Balanced, Flexibility.

Multi-system execution with referential integrity

Enterprise Platform coordinates Oracle, MongoDB, and Kafka in one run. Deterministic end-to-end, with per-stage execution logging.

Template engine for industry messages

Spec-aware form editors for EDIFACT, SWIFT MT, HL7 v2.x, and HL7 FHIR. Upload your own specs, validate in real time, generate deterministic test messages.

SWIFT messages are test-only artifacts

Generated messages are syntactically valid but not network-validated. They must not be transmitted on production SWIFTNet or from a CSP-attested secure zone.

Governance layer for regulated enterprises

Role-based access, audit dashboards, approval flows, reusable templates, and scheduler. Deployed in EU banking for Oracle, MongoDB, and Kafka pipelines.

Compliance evidence, not attestation

DATAMIMIC produces reproducible artifacts for GDPR, DORA, PCI DSS, and HIPAA programs. It does not replace your DPO, CISO, or auditor.

CE and EE are separate engines

They share the DSL and determinism contract. CE is Python with multiprocessing. EE is independently optimised with Rust, ML, and distributed execution.

Seven domain packs in Community Edition

Healthcare, Finance, Insurance, E-commerce, Public sector, Demographics, and Common. All services are versioned, seeded, and emit provenance hashes.

CLI is the baseline agent contract

Capabilities returns the structural surface. Reference queries load typed fragments. Scaffold compiles intent to XML. Lint and dry-run verify before real execution.

MIT licensed, contributions welcome

Community Edition is open source. If you are building integrations, domain extensions, or MCP tooling on top of DATAMIMIC, they want to hear from you.

What to remember

Same seed plus same model always produces byte-identical output, making test data reproducible across runs and machines.

Read the original on GitHub

React

Sign in to react and comment.

Comments (0)

Life is short. Keep it sweet. Respect others' opinions and be kind!

    Recommended next

    More decks on datamimic and related topics.