About the Builder

Maryam Moadeli

AI Systems Architect & Technical Product Lead

Building thoughtful AI systems, writing honest field notes, and treating ambition like an endurance sport.

I build systems that are useful in real life, not just impressive in demos.

At Google, I operate across systems architecture and technical product management, sitting right in the middle between engineering teams and business leadership. I like being the person between big ideas and working software: the one asking, “what problem are we actually solving?” before the architecture diagram gets too confident.

What I Care About

Most agent demos are impressive once and unreliable forever. I spend my time on the less glamorous parts: deterministic tool validation, strict schemas, and error recovery that assumes things will break, because they usually do.

My sweet spot is making complex systems understandable, scalable, and calm under pressure.

What I’m Building

My main personal testbed is Moka, an agent system exploring persistent memory, scheduling, and context that doesn’t evaporate the second you close the tab. Parts of that work are open source in NanoClaw, while the fuller setup runs quietly in Docker, occasionally humbling me.

This blog is my lab notebook. I write about what I shipped, what broke, what surprised me, and what I learned five minutes after thinking, “this should be easy.”

Advisory & Mentorship

I love translating sharp technical work into language that makes people feel smarter, not smaller:

• Google for Startups Accelerator: Technical mentor helping founders figure out where agents create real leverage versus where they’re just expensive theater.
• ADPList: 1,200+ minutes mentoring engineers, PMs, and architects. Long enough that I’ve started catching myself repeating my own advice.

Outside The Repo

I spend a lot of time reading about cognitive psychology. Turns out “how humans actually think” is fairly relevant when you’re building software that’s supposed to reason on its own. Half of debugging an agent is realizing it’s making the exact same reasoning mistakes people do, just ten times faster and with absolute confidence.

I also run distance. It has the same feedback loop as systems and product work: track the signals, ignore the noise, adjust the pace, and deal with the occasional knee that completely disagrees with the plan.

If you want to talk agents, systems, product thinking, or running, say hi at mindmoation@hexora.ca.