Your CTV Metadata Is the Model's Blind Spot
CTV is adopting outcome-optimization models faster than it is fixing the content signals those models are supposed to learn from. The result won't show up...
7 min read
Fourteen years keeping enterprise systems honest taught me how software fails. Now I build AI systems, web platforms, and a native iPhone app that don't.
My career runs from a technical support desk at Verint to the streaming ad pipelines at FreeWheel, with national-scale video infrastructure at Comcast and Charter in between. At FreeWheel I led cross-functional investigations into complex streaming failures, built SSAI diagnostic tooling the broader engineering team came to rely on, automated ad-inventory reporting that had been hours of manual spreadsheet work, and ran the stakeholder syncs that kept engineers and executives on the same page. You learn a particular honesty working on systems that millions of people are watching.
In 2025 I went independent, and the way I work came with me: documentation before conversation, contracts before code, and failure modes named before they happen. I build agentic AI systems that connect language models to live data and real tools, and I run them in production — not demos, infrastructure. The detailed AI work lives at lexg.ai.
The last year is the proof I point to: I picked up Swift with no iOS background and shipped the iPhone app my booking business runs on — learning a new platform the same way I learned every other one, by building something with real stakes and letting production grade the work.
I live in Denver. I studied psychology before I ever touched a server, and it still shows up in everything I build — software gets used by people under pressure, and designing for how they actually think is half the engineering.
2025 — Present
Independent — LexG.ai & Gaines Concierge
Agentic AI platforms, production MCP tooling, and a native iPhone app running a real booking business.
2018 — 2025
FreeWheel, A Comcast Company — Senior Systems Engineer
Streaming ad-tech (SSAI) at national scale: incident investigations, diagnostic tooling, reporting automation, stakeholder communication.
2016 — 2018
Charter Communications — Advanced Video Engineer
Data pipelines and internal web tools supporting the Spectrum Guide rollout across national headend infrastructure.
2011 — 2016
Comcast — Network & Advanced Video Support
National-scale Linux infrastructure, automation scripting, and cross-team technical point-of-contact work.
2007 — 2009
Verint Video Solutions — Technical Support
Where the habit started: listening to what's actually wrong before deciding what to fix.
The AI lab: agentic chat with live tool orchestration, dual-backend inference, voice dictation, and the fleet of MCP servers behind it all.
A client booking business run from a native SwiftUI iPhone app I built end to end — verified in production, in daily use.
Working ideas out in public.
CTV is adopting outcome-optimization models faster than it is fixing the content signals those models are supposed to learn from. The result won't show up...
7 min read
What this small LangGraph starter leaves out, a concrete list of next steps, and what I learned and would do differently.
4 min read
Every failure from building a local LangGraph agent: a stale ChromeDriver that made the model say a site was down, invented dates, and more.
10 min read