About Us

About Us – Gemini Build Lab

I am Leo Whitaker. I live in San Mateo, California, and work as an AI product and growth operator at a Bay Area startup. I started as a software engineer after studying computer science, spending years writing Python, building APIs, wiring data pipelines, and shipping early LLM features before moving into product, growth, and operations. Today I use Gemini, ChatGPT, and Claude every day. I only write about a feature after I have tested it myself—from a clean account or a fresh project whenever possible.

The site exists because documentation tells you what a product can do. I want to show what actually worked, what failed, why it failed, and how to build the smallest useful version yourself.

From Engineer to Operator Who Still Opens the Terminal

My early career was hands-on engineering: Python clients, authentication flows, data transformations, and the quiet work of making systems reliable enough for other people to depend on. That foundation still shapes how I evaluate new models. A polished demo is easy. The edge cases, the retry logic, the cost surprises, and the moments when the model drifts under real load are the real tutorial.

At the startup I help decide where Gemini fits in our stack and where it does not. I participate in those conversations, but I never publish company secrets or internal roadmaps. What I do publish are the experiments I can reproduce outside any proprietary environment: API clients, structured-output patterns, function-calling setups, simple RAG apps, and the workflows I actually run during a normal workday.

Evenings, Road Trips, and Clean-Account Tests

I am single and stay connected through AI meetups, developer events, and introductions from friends. Evenings and weekends often go to writing, coding small experiments, and keeping tutorial repositories current. I enjoy photography, short video shoots, and road trips around California and the American West. Those trips force me to pack light and decide what is actually useful—the same discipline I try to apply to AI tools.

Before I recommend a workflow, I try to recreate it from a clean account or a brand-new project. If it only works inside my already-configured environment, it is not ready to share. That habit has saved me from publishing more than one “it worked for me” claim that would have failed for a reader starting from zero.

I spend carefully on everyday purchases but willingly pay for tools, courses, and travel that are genuinely useful. The same filter applies to the models and libraries I write about: I only recommend what I would still use after the novelty wears off.

Why Gemini Build Lab

I write for developers, technical product managers, data scientists, automation builders, and startup teams who want to move past simple chat prompts. The site sits between beginner tutorials and dense official documentation. Every major post aims to give you the smallest working example first, then the architecture notes, the limitations, the costs, and the failure modes.

The four sections—Gemini Fundamentals, Build with the API, Workflow Lab, and Field Notes—mirror how I actually learn a new capability: understand the model, build the client, put it into a real workflow, and write down what broke. The signature rule is simple: I tried it first. You can just copy.

I do not write about features I have not used. I do not promise instant wealth or effortless automation. I do not turn model comparisons into tribal arguments. I label versions, dates, environments, and known limits whenever they matter. If a workflow cannot survive a clean-account test and a Monday-morning re-run, it does not belong here.

If you are looking for hype or secret prompts, this is not the place. If you want the practical version that survived real testing, you are closer to the right neighborhood.

— Leo Whitaker

Gemini Build Lab

San Mateo, California

Last updated: