New York City

Jonathan Cruzate.

Technical Program Manager at Google · PMP® · New York City

Experience

Google — Core Data, Machine Learning & Infrastructure

New York, NY · November 2025 – Present

Co-lead multi-year compute, storage, and accelerator efficiency strategy across large-scale cloud and AI/ML infrastructure, driving 4–5% fleet-wide resource-efficiency targets with 80%+ high-confidence attribution coverage.
Architect automated capacity-governance workflows and analytical models that reconcile financial spending plans with physical cluster allocations.
Design and deploy production multi-agent AI orchestration systems, background event-stream daemons, and automated quality gates.
Drive technical program management across AI/ML, personalization, serving, and infrastructure-efficiency workstreams, with standardized RAID tracking and automated executive briefings.
Guide delivery of modular project-chartering, automated issue-tree health-monitoring, and continuous documentation-synchronization tools.
Mentor TPM interns and cross-functional contributors from system design through production launch.

Google

New York, NY · April 2025 – August 2025

Directed a search indexing modernization program for Google Drive — an 8,000+ engineering-hour initiative migrating legacy systems to a unified, sharded ST-Spanner backend.
Resolved a critical performance bottleneck by identifying resource contention, coordinating additional compute capacity, and eliminating service throttling.
Launched a unified request management system for a core authorization service, streamlining feature intake from Google Drive, YouTube, and other major teams.
Authored the official project management framework for the 100+ employee Storage Ecosystem, standardizing planning, estimation, and OKR alignment.

Open Source — Python

Command-line interface for Cronometer: diary logging, food search, nutrient targets, and biometrics from the terminal.
Reverse-engineered the Android app's private mobile API (Dart AOT binary reconstruction) to recover its token-based auth flow after no public API existed.
Date/time-pinned, read-back-verified writes with dry-run mode and a full offline test suite.

Open Source — Python

One search across local files, Google Drive, OneDrive, Dropbox, Google Keep, and email headers.
Prebuilt local indexes (~240k records) answer queries in about a second with zero model calls and no network — a deterministic script, not an LLM.
Watermark-incremental index refresh with a live CLI fallback on index misses.

Open Source — Python

Smart-home automation patterns for Muse users, generalized from a real working home: one automation engine driving home, away, and night modes.
Three-source presence voting, spoken-alert routing over Google Home, SmartThings device control, and a universal MacroDroid macro for phone control.
Published as a fully scrubbed edition — every example device, name, and value is a fictional placeholder.

For work inquiries, collaborations, or anything else — send a note. It goes straight to my inbox.