I build production-grade automations that cut LLM costs by 80% — and actually handle the edge cases that break demos.
Software engineer by trade — AI automation by design. n8n + custom code, deterministic scoring over guesswork, human-gated AI pipelines.
Software engineer by trade — I build AI automation the way real software gets built: tested, traceable, and debuggable when the no-code path hits its limits. I work with founders who need automation that holds up over time, not a demo that breaks on the second real input.
Most automators either know the tools or know the engineering. The combination is rarer, and it's what keeps a workflow running when traffic spikes, edge cases appear, or an API silently changes. By day I'm a Senior Software Engineer at Everyone Technologies — promoted January 2026 — shipping multi-tenant SaaS to hotels, distributors, restaurants, schools, and LGU clients in the Philippines. On the side I build n8n and custom-code systems that hold up over time, including an MCP server I run daily to keep my own projects documented. I'm also building Kuwenta, an offline AI finance tracker I'm launching on the Play Store — because some jobs are better handled by a small local model than a cloud API call.
I don't just configure automation tools — I build the logic behind them. Prompt engineering, branching workflows, error handling, memory compression, fallback responses. The systems I build are production-grade because I understand what breaks when real traffic hits, and I build for that upfront.
The stack behind every automation
Not limited to these — if your tools have an API, I can integrate with them and go deep.
I don't just build automations. I deploy and host them — you get a live, running system, not a workflow file.
I work with businesses drowning in manual work — and build the automations and software that make it stop.
Replying to the same customer questions, copy-pasting data, sending the same weekly report manually?
I automate the repetitive parts of your operation — customer replies, escalations, reports, cleanups — so your team focuses on what actually matters.
Your automation hits a wall — there's no platform for it to talk to, no portal for your customers, no single source of truth for your data.
I build the custom system that serves as the backbone — orders, inventory, customer data — so your automations have something real to work with and your team has one place to look.
Not sure how to get your software out of development and into the real world — or how to keep it running?
I handle servers, domains, SSL, backups, and monitoring. You get a real URL that works for your team and your customers from day one.
Six systems running in production — click through the diagrams to see how each one actually works.
AI lead scoring into self-hosted booking — end to end
Click any node to see what it does.
Overview
The whole system end to end — a form submission becomes a scored lead, a tier-matched reply, a self-served booking and a reminder. Each later tab zooms into one stage of this.
Why
Leads from my contact form sat in my inbox for hours while I was deep in client work. By the time I replied, they had cooled — and the ones worth replying to first were indistinguishable from the ones that weren't.
What
Scores every inbound lead on a transparent 100-point model, writes a reply matched to that score, and hands the best ones a booking page that reads my real calendar — no third-party scheduler in the loop.
How
Two separate AI passes read the message: one extracts qualification signals, the other writes prose. Between them sits a deterministic weighted score, so the tier is decided by auditable arithmetic rather than by the model's mood. Hot leads get an immediate reply carrying a signed booking link; warm and cold become drafts I approve. The booking pages are served by the workflow itself.
Stack
Straight answers about scope, cost, timelines, and ownership.
I design and build production systems that do repetitive work for you: lead handling, customer replies, data cleanup, reporting. Not demos — systems that keep running when real traffic and edge cases hit, built with real engineering (error handling, retries, logging), mostly on n8n plus custom code.
If it has an API, yes. Common ones: Gmail, Slack, Telegram, Notion, Cal.com, Google Sheets, HubSpot, Postgres. I go deep on integrations rather than settling for the happy path.
It depends on scope. Start with a free process audit — tell me one thing your team does manually and I'll tell you what's automatable and roughly what it takes. No obligation.
A focused automation is usually days, not months. Larger multi-system builds are scoped up front with weekly updates so you always know where things stand.
You own everything. Full source, handover documentation, no vendor lock-in. If we stop working together, nothing stops running.
Yes — remote worldwide, async-first. I'm based in the Philippines and reply within 24 hours.
Send a message — it goes straight through a live AI scoring pipeline. Hot leads get a response first.
AI automation pipelines, business systems, and deployment — end to end.
Weekly updates
You always know where things stand
You own the code
Full source, no lock-in, ever
Handover docs
Every project includes documentation
Async-first
PHT evenings, replies within 24hrs