I design and build cloud systems and AI applications for enterprise. My background spans architecture, hands-on engineering, and the messy human work of getting technology adopted at scale.
Currently I'm focused on agentic systems, and how they help organizations turn complex technology into real business value.
Coding agents get resourceful when something is missing, and prompts are not boundaries. Here is how I give each agent its own identity and container, and map it into the controls we already have.
I designed and wrote my GPG and YubiKey guide with two AI models. I expected it to be faster. It was not. Here is what AI actually changed about how I work, and where my own discipline still mattered.
Building AI applications end to end, from agentic pipelines to the UX
I'm designing and building a Kubernetes-based platform where autonomous coding agents turn GitHub issues into pull requests. I scope the issues on my laptop with a coding agent. Then each issue runs in its own disposable container, with its own GitHub account @julieio-agent and no access to my credentials or production. Nobody watches the session, and I review what comes back.
I built an AI app that turns a hand-annotated receipt into a fair split. Underneath the simple idea: a multi-layer LLM pipeline, computer vision for the handwriting, and a custom HMAC security model.
An AI advisor for designing Azure Kubernetes clusters. It blends official Well-Architected best practices and Microsoft docs with guidance personalized to your requirements. I aimed to digitize the version of me that only Microsoft's enterprise clients had access to.
One app is easy to automate; add components, infrastructure and teams and you're lost. How many pipelines and repos do you need? Criteria to slice your pipelines without spaghettifying them.
Monitoring tools drown you in information. Building a microservice architecture visualization with d3.js and live monitoring, then watching your architecture dance in real time.
JavaScript's quirks make CI/CD tricky. What do you compile, when, and where? A look at full-stack JS pipelines and the common challenges. We'll laugh and cry.
