I'm a full-stack engineer who does machine learning properly, which means I can take an idea from a notebook to something people actually use: the model, the API around it, the web and mobile apps on top, and the cloud infrastructure underneath.
My ML grounding runs from logistic regression through to transformers, with convolutional networks for computer vision along the way. During the Covid lockdowns I built an AI symptom checker and report builder. Today that work lives on in Harmoni Hub's AI agents and smile preview tool, running on my own open-source agent library, tenxagent.
The product side is just as hands-on. Back ends in PHP with Laravel, Python with FastAPI, and Node. Front ends in Vue and React, and mobile apps in React Native: the Harmoni staff app for admin, compliance and HR, with a patient app on the way. I deploy to AWS (EC2, Lambda and container images in ECR) and DigitalOcean, so a model never stops at a demo.
I test like it matters, because in healthcare it does. Laravel projects ship with PHPUnit feature suites, Mockery and Laravel Dusk browser tests, kept tidy with Pint and run locally on Sail. Python services run on pytest, with pytest-asyncio for async APIs, httpx for endpoint tests, moto to stand in for AWS, and mypy for type checks.
Now I build with AI agents every day. The years before that are why it works: I know what a sound data model looks like, where systems break, and when generated code is confidently wrong. Agentic coding makes me fast. Experience makes the result robust.