I'm an AI engineer in Metro Manila. I build agents, MCP servers, and the automation around them, I design the interfaces people run them from, and I build with AI the whole way down.
Claude Code is how I write, not a tool I reach for now and then. Most days I'm taking a fuzzy request, scoping it into something an agent can actually do, wiring it to real systems, then watching what it does under load and fixing what breaks. Scope, build, integrate, measure, repeat.
The part I care about is what happens when an agent is wrong. So I don't assert my systems are safe, I measure them: I run the same agent ungoverned as a control and let the numbers make the case. I put a person at every step where a decision has consequences, and I report the cost and the failure modes, not only the wins.
I work in Python and TypeScript, against Claude and OpenAI, with RAG over real corpora, orchestration in LangGraph and hand-written tool loops, structured outputs and tool calling, and evaluation pipelines that track quality, reliability, and cost, traced per call in Langfuse. The interfaces are React and Next.js, designed in Figma first. At scale that has meant agent fleets running unattended for weeks, retrieval over thousands of chunks of law, and a content pipeline that produced roughly 300 phishing simulations across eight-plus markets and four difficulty tiers.
My path here was not a straight engineering line, and that is the point. Electronics engineering degree, then an associate software engineer seat at Accenture, then design, then design engineering at a global cybersecurity company where I became a one-person production line for AI tooling and simulation content. Each step left something: how systems get tested, how interfaces get built, how work really moves through a business. The agents I build now sit on all of it.
AI Engineer, AI Agent Engineer, Design Engineer on AI products, Founding Engineer, and full-stack roles at product companies that ship with AI. Remote, and used to a wide timezone offset. If you are hiring for someone to own agents end to end, scope them, build them, integrate them with real systems, measure them, and keep them safe to run, that is the work I want.
"What makes her stand out is that she doesn't just appreciate great design. She knows how to adapt it to the project and actually ship it. She can develop and vibe code her own work, which means things move faster and the gap between design and execution is much smaller."
Erlich Maramot · Creative Director & UX Strategist, 15 years
"She helped us explore AI automation workflows, making complex concepts accessible and actionable for the team. She has a clear ability to bridge the technical and the practical, which is not easy to find."
Ali Abbasov · Cybersecurity Specialist, worked with her at the security company
"She takes ownership of her tasks and ensures that every deliverable meets a high standard."
Carmeli Marie Joy Ancan · managed her directly
Verbatim from LinkedIn recommendations ↗