Appearance
Engineering posts on identity resolution, data architecture, and building with Kanoniv.
The identity fragmentation problem: when agents use different identifiers for the same person, memory fragments and coordination fails. A technical walkthrough of the problem and the architecture pattern that solves it.
Entity resolution from a data engineering perspective: what it is, how it works (blocking, matching, clustering, survivorship), what tools are available, and how to implement it in production pipelines.
A feature-by-feature comparison of Kanoniv, Splink, Zingg, Dedupe, and Senzing. Matching approach, golden record support, scalability, licensing, and when to use each tool.
How to process millions of records efficiently: blocking strategies, parallelization, incremental matching, and benchmarks. Includes results showing 1M records processed in 95 seconds.
A technical deep-dive into Kanoniv's reconciliation engine: why Rust over Python, the PyO3 bridge, parallel comparison with Rayon, and performance benchmarks showing 100x speedup over pure Python.
How to match patient records across EHRs, claims systems, and labs while maintaining HIPAA compliance. Matching strategies, PII masking, audit requirements, and common pitfalls.
Why modern data teams are choosing declarative, spec-as-code identity resolution over monolithic MDM platforms. Cost, time to value, and integration with the modern data stack.
A step-by-step tutorial: raw data in Snowflake, staging models in dbt, cloud reconciliation with Kanoniv, and reverse ETL back to the warehouse. 1.58M records resolved into 612K canonical customers.
