I started with networks.
I'm a software engineer in Manila. I work on AI systems for enterprises — regulated data, legacy systems of record, networks that block everything until someone says otherwise.
I didn't start there.
I started with networks: switching, routing, the parts of computing you can point at. Then site reliability, watching microservices on Kubernetes and writing automation to catch errors before customers did. That was the first time I saw production when it's unhappy, which turns out to be the most useful thing I've learned.
Then four years in insurance, on unglamorous problems. A notification service taking thirty seconds and failing a third of the time. 1.6 million records that had to move without anyone noticing. An infrastructure bill twice what it needed to be. Features across four APAC markets, where coordination turned out to be harder than code.
Now I lead engineers building agentic AI for enterprise clients, and teams that build with agents — in 2026 I gave the plenary at Globe DevCon on taking a platform from zero to first release in 108 days with an AI teammate. Same job, really. One puts the model inside the product, the other inside the process, and both only work if you're honest about what it can be trusted with.
How I work
Start at the fundamentals. I'd rather understand why something behaves the way it does than collect workarounds for it. The notification service that took thirty seconds wasn't a slow-code problem — it was transaction scope and indexing underneath, and three separate symptoms turned out to be one cause. You only see that if you go down a layer.
Quality comes from iterating and testing, not from a final review. I work in small changes, each one tested before the next. Load and failure scenarios get simulated before anything reaches production, and tests live in the pipeline rather than in someone's memory. It's also what keeps a rollback cheap.
Leverage agents — for my workflow and the team's. I build with AI agents every day, and I set up how teams around me use them. Done properly it moves a whole organisation, not one engineer. Done carelessly it ships work nobody actually checked, so the rule stays the same: the agent drafts, a human decides.
Design for scale and reliability from the start. A system has to hold up when everything is happening at once, not when nothing is. Most of my four years in insurance went on exactly that — contention, latency, live migrations, and arranging things so that going wrong costs a code revert instead of a weekend.
Also
I run skills-development programmes for engineers, which is the part of the job I'd keep if I had to give up the rest. Studying Computer Engineering at the Technological Institute of the Philippines, finishing 2027.
Certifications
The stack
Java and Spring, Python, Node, Angular. Azure and AWS, Kubernetes and OpenShift, Docker, Terraform. MongoDB, Redis, MySQL, Azure SQL. New Relic.
Working on the same thing?
If you're working through any of this, I'd like to hear about it.