Peregryn · Methodology, Security & Data Residency

Your data never leaves. Neither does the decision trail.

Peregryn runs entirely inside your own environment on commodity x86 — no GPU, no cloud service anywhere in the path from profile to merge. Every match is a scored, reproducible decision made where your data already lives.

Posture at a glance

Built for a controlled local deployment, not a hosted matching service.

The full pipeline — profile, cleanse, standardize, match, merge — executes locally inside the customer-controlled deployment environment. Records are not sent to an external service for matching.

Runs where

Inside your environment, against data already staged there. All five pipeline stages execute locally.

Data movement

None outbound. Records, match scores, and merge decisions stay inside your environment. No cloud services are used.

Method

Deterministic rules plus Fellegi-Sunter-class probabilistic scoring — a published statistical method your model-risk and governance teams can review, not an opaque black box.

Reversibility

Merge and unmerge are first-class operations. Any merged cluster can be rolled back in full.

Hardware

Commodity x86. No GPU, no accelerator, no minimum cluster size.

Throughput

12 million records matched end-to-end in under 7 minutes, measured. Reproduced on your data during a proof-of-value engagement.

Change control

Match rules and survivorship remain transparent and controlled.

Matching thresholds and survivorship rules are explicit configuration, reviewable by your data-governance function before anything merges. Nothing merges on defaults you have not seen — because an engine that can merge records should never guess at intent.

Scope we state plainly

Peregryn is a data-processing engine deployed inside your environment; it inherits the certifications and controls of the environment you run it in rather than carrying its own external attestations. PDI works with your security team to place it within your existing compliance boundary. Specific control mappings are provided during scoping.

PIPELINE · PROFILE → CLEANSE → STANDARDIZE → MATCH → MERGE · DAMA-DMBOK-ALIGNED

Methodology review

Review the methodology with our team.

PDI will walk your data-governance, model-risk, and security stakeholders through the matching method, survivorship rules, rollback model, and deployment posture against your environment.