- Built the internal operations platform (Next.js and PostgreSQL) in ten weeks.
- Built an unattended LLM extraction pipeline that turns unstructured meeting transcripts into a structured signals database.
- Killed my own flagship predictive-model project when a label audit showed the data could not support an honest model.
- Every deploy backs up production, restores it into a scratch database and diffs the row counts before anything migrates.
- Access control rebuilt from 55 scattered permissions into a six-area grid with no bypass role by design.
- Its own gate suite of 300+ checks wired into the build.
Full Stack AI Engineer
Hand me a goal instead of a spec and I come back with the diagnosis, the recommendation and the system.
Cebu City, Philippines, remote
- checks in the build gate
- 300+
- checks in the build gate
- to build the operations platform
- 10 wk
- to build the operations platform
- permissions rebuilt into six areas
- 55 to 6
- permissions rebuilt into six areas
Ask about Mark
Ask it something, and if the answer is on the page it will take you to it.
How I work

- What I ownAI and data automation at WeAssist, end to end.
- What arrivedA goal rather than a backlog.
- What I built firstThe picture of how work really moved between departments and the tools they already had, written up with a buy-versus-build recommendation.
Work
Hatchit Solutions
Software Engineer
June 2025 to April 2026
Core systems for a multi-tenant ERP platform.
Hatchit Solutions
Web Engineer Intern
January 2025 to May 2025
Full project lifecycle across design, development, testing and deployment.
Things I built to be inspected
ScoreAudit
Not taking a model's word for it.
Ask a language model six questions with checkable answers and make it commit to how sure it is.
- about 5 to 12 seconds
- runs for real when you press the button
Split-Brain Sandbox
Making sure exactly one worker owns a job when the network cannot be trusted.
Three nodes, one job, and only one may run it.
- about 6 to 14 seconds
- runs for real when you press the button
Open source
public-agent-provisioning
A forkable template for making AI coding agents safe by construction.
- Rules loaded every turn.
- Skills loaded on demand.
- Hooks that intercept tool calls before they run.
- Git guards that block the commit.
- A self-check tier that forces every guard to fire and fails if it does not.
Stack
- TypeScriptJavaScript with static types, checked before the code runs.
- Next.jsThe React framework handling routing, rendering and builds.
- Node.jsJavaScript on the server, outside a browser.
- PostgreSQLThe relational database most of this work is built on.
- PythonGeneral-purpose language, the default for data work and scripting.
- Claude CodeAnthropic's coding agent. Runs in the terminal and edits real repositories.
- MCPAn open standard for giving a model tools and data sources.
- Cloudflare WorkersCode that runs at the edge, close to the visitor. This site is one.
+ 47 more tools- hide the full list
- Languages
- JavaScript, SQL, Bash, HTML, CSS
- Frontend
- React, Tailwind CSS, shadcn/ui, HTMX, Figma, Motion
- Backend
- Bun, NestJS, REST APIs, Zod, BullMQ
- Data
- SQLite, Redis, Drizzle ORM, Prisma ORM, Cloudflare D1, NocoDB
- AI and agents
- OpenRouter, Model cascades, Vercel AI SDK, RAG, OpenCode, Antigravity, Codex, Whisper STT, ElevenLabs, Kokoro TTS, Pinecone
- Infrastructure
- Docker, Docker Compose, Cloudflare Tunnels, Coolify, Hetzner, Caddy, Tailscale, Linux, Git, CI/CD
- Automation
- n8n, GoHighLevel, Webhooks, Cron
Education
BS Information Technology, Magna Cum LaudeCebu Institute of Technology - University2021 to 2025- AWS Certified AI Practitioner (AIF-C01), 2026, Credential ID AWS06249557
- TOPCIT Level 3
- PhilNITS Fundamental Engineer (FE)