Hey, I'm

Michael John Angelo Belciña

I build backend-heavy systems, crawlers, automation pipelines, and SEO tools. Lately I've been deep into agentic AI workflows and shipping with coding agents.

Currently: building SEO audit modules at Poliris.io, shipping FastAPI APIs and extending automated audit tests.

Get in touch GitHub LinkedIn Resume
Available for work
About

Full-stack developer from Davao City, Philippines with a strong focus on Python APIs, data processing, and workflow automation. I've shipped production SEO platforms, scraping pipelines, and automation workflows. I'm actively exploring agentic AI, building with LLMs, and using coding agents like Claude Code, Codex, and OpenCode as core parts of my development workflow. Open to diving deeper into agentic work and AI-driven tooling.

Languages

  • Python
  • JavaScript
  • TypeScript

Backend

  • FastAPI
  • Hono
  • REST APIs
  • Celery

Data & Infra

  • PostgreSQL
  • Redis
  • Supabase

Automation

  • Web scraping
  • n8n
  • SERP extraction

Frontend

  • React
  • Next.js
  • Vanilla JS

AI & Agents

  • OpenRouter
  • Claude Code
  • Codex
  • OpenCode
Work

Full-Stack Developer

Poliris.io · Contract · Remote

Collaborating with an SEO domain expert to build core product modules end-to-end across backend and frontend. Building FastAPI APIs with cursor-based pagination, response caching, OAuth integration, and token encryption, and shipping Next.js dashboards with paginated tables, state persistence, and deep-linked navigation. Maintaining and extending 50+ automated audit test modules with partial scoring, batch orchestration, and detailed failure reporting while improving performance through bounded concurrency and batched API calls.

Python FastAPI Celery Redis Supabase Next.js

Automation & Python Developer

Contract · Remote

Built SERP scraping pipelines, processed keyword-based lead data with rotating proxies, and developed automation workflows for outreach and content processing. Created an n8n workflow to turn blog posts into weekly newsletter drafts using LLM summarization.

Python TypeScript n8n Serper.dev

Education

BS Computer Science, University of Mindanao · December 2025

Projects

Bifur

Private

A multi-agent social simulation engine built with TypeScript and Bun. LLM-powered agents autonomously generate posts, replies, and arguments in threaded conversations anchored by real-world news and persistent societal issues. Each agent has its own belief system that shifts through debate. Uses OpenRouter for LLM inference, with configurable tick rates, participant pools, and conversation triggers. Includes a real-time web UI to observe agent behavior, start/stop simulations, and inspect belief changes live.

TypeScript Bun OpenRouter LLM Agents Multi-Agent Simulation

Spec

View on GitHub

Turns human-written or LLM-written Markdown into executable browser tests. Write test plans in plain Markdown with steps and expectations, and Spec compiles them into a strict runtime format, runs them through Bun with real browser automation, and saves results as artifacts.

TypeScript Bun Browser Testing LLM

Sik

View on GitHub

An API-first platform for SERP scraping and workflow automation using alternative search engines. Lightweight request-based scrapers without browser automation, built with Hono and Bun. Integrates with n8n for automated pipelines and supports CSV/Google Sheets export.

TypeScript Bun Hono SERP Scraping n8n