Drift Intel
$3M+ saved / yearReplaced a rigid, single-vendor data pipeline with a modular, plugin-based architecture. New data-partner integrations dropped from three months to about a week — unlocking cheaper partners and over $3M in annual savings.
Customer Sophistication Model
NDR 79% → 121%Built a tiered model linking product usage to outcomes. Top-tier customers saw 4.8× daily active conversations and 9.5× pipeline; it became core to how Customer Success onboards and evaluates accounts.
Contacts at scale
millions → ~20BDesigned the systems behind Drift's contact graph as it grew from millions to ~20 billion records and 1 TB to 90 TB. Polyglot persistence across MySQL, DynamoDB, Elasticsearch, and S3 — 31M daily visits at 15K OPS.
The 64-bit migration
0 downtimeContacts were capped by 32-bit IDs — a looming Y2K-style shutdown. Led a 12-person team rewriting ~10% of the codebase across 600+ PRs, moving primary keys to 64-bit with under five minutes of downtime.