LinkedIn slashes data traffic 99% with homomorphic encryption

LinkedIn revamped its Revenue Attribution Report pipeline by applying additive symmetric homomorphic encryption, letting it aggregate encrypted ad data directly in Apache Pinot. The trick cuts network traffic by more than 99% and keeps member data private, opening the door for large‑scale privacy‑preserving analytics.

Apache Parquet 1.15.0 and earlier contain a critical schema parsing flaw (CVE-2025-30065) that lets attackers execute code via crafted Parquet files. The issue scores a perfect CVSS 10.0, and the fix arrives in version 1.15.1. Pipelines that ingest untrusted Parquet data must upgrade immediately.

Zillow built a knowledge graph that stitches together listings, neighborhoods, schools and other home data into a unified network. The graph lets them normalize vocabularies, enrich unstructured text and power search features that surface more relevant homes. The post walks through ontology design, extraction pipelines and real‑world product use cases.

Zillow built a knowledge graph that stitches together listings, neighborhoods, schools and other home data into a unified network. The graph lets them normalize vocabularies, enrich unstructured text and power search features that surface more relevant homes. The post walks through ontology design, extraction pipelines and real‑world product use cases.