Senior Director of Engineering · Two-time founder
Gary Pezza
I build and scale engineering organizations, and I still write code most nights and weekends. Lately that means iOS apps, AI tools, and games.
About
Leader by day, builder by night.
I'm a Senior Director of Engineering at SimpliSafe, where over the past several years I grew my organization from 4 to 30 engineers across 5 countries, restructured it into six independent squads, and helped drive our app from a 1.5-star rating to #1 in its category at 4.8 stars, now serving 1.3M+ monthly active users.
Before that I was VP of Engineering at GTxcel and co-founded Requiem Software Labs, where we shipped 40+ games and apps to 4M+ downloads. Today I'm also co-founding Zoned Labs, an AI-powered recruiting platform for college athletes.
Across twenty years the work has stayed the same: building things that ship and staying close to the code.
Experience
Where I've built and led.
Senior Director of Engineering · SimpliSafe
Grew the org from 4 to 30 across 5 countries and six squads. Drove the app from 1.5★ to #1 in category (4.8★), incubated the AI team, and shipped the company's first outdoor and indoor cameras.
Co-Founder · Zoned Labs
AI-powered SaaS that matches college athletes with realistic recruiting targets. Built end-to-end: React/TypeScript, Node, an AI coach with function calling, multi-tenant white-label architecture.
VP of Engineering · GTxcel
Led a 20+ person engineering and product org. Built a headless CMS powering ~300 publishers, migrated infrastructure to AWS (50% cost cut), and pivoted the platform into a new vertical in four months.
Co-Founder · Requiem Software Labs
Shipped 40+ games and apps to 4M+ downloads and $3.75M in revenue. Built an in-house cross-platform Java game engine and ran multiplayer infrastructure for 100K+ concurrent users.
Senior Software Engineer · Raytheon
Led the infrastructure team on the $2B Zumwalt-class destroyer program, building common C++ tooling to integrate simulation systems. Three performance awards.
Building with AI
I instrument my own AI coding and tune the setup with real evals. Most of my building now happens alongside coding agents, and to get better at it I treat my own workflow as something to measure and improve, not just a habit. I log every Claude Code and Codex session into a database and study how I actually work with AI.
Measuring how I direct the work
One report classifies every prompt I write (planning, course-correction, redirect, learning) to separate the output I actively directed from what the model produced on its own. It turns a vague sense of "vibe coding" into numbers I can track over time.
Tuning the setup with evals, not vibes
To improve my context, skills, and tools, I run an offline evaluation loop: I hand-review conversations as ground truth, then score prompt and model versions against those labels so I can compare changes honestly instead of by feel.
Still building
Some of what I'm building now.
iOS apps
Native Swift and SwiftUI apps for the App Store: countdown timers, road-trip games, and small utilities for my family.
AI tools
LLM-powered agents and products: a WordPress AI agent, recruiting coaches with function calling, home-improvement estimators.
Games
Building in Godot and GDScript: arcade shooters and experiments, picking up where the Requiem game engine left off.
Sites and web apps
Full-stack work in Next.js and React, from simple personal sites like this one to a fielded estimating tool with a custom pricing engine, accounts, and PDF quotes.
Apps you can use
Shipped and live on the App Store.
