01 / IndexAI Systems EngineerAI Systems Engineer · Technical Operations

Open to rolesOpen to AI systems, backend and platform rolesTacurong, PH · UTC+08:00

MarkAnthonyResoso.

— Now

Runs production voice & messaging AIfor 12–15 live businessesacross Ireland and the UK.

Owns the stack from the voice layer down: agents, orchestration, database, integrations, infrastructure and monitoring.

live clients across Ireland & the UK: 12–15
booking systems integrated, most reverse-engineered: 5
faster database after a zero-downtime EU migration: 12×
production findings documented with root cause & proof: 129

I build and run AI systems in production: ElevenLabs voice agents, n8n orchestration, Supabase Postgres, Railway, and the monitoring that keeps them up. Before that I shipped full-stack Laravel, Express and React systems for government, agribusiness and retail clients.

View work Get in touch

— Currently operating

AI receptionists for 12–15 businesses across Ireland & the UK — from the voice layer down to the database.

Two front doors, one brain. The phone line and the DM inboxes call the same tools, obey the same rules and write to the same records.

Voice
CallerTwilioclient numberElevenLabs agentprompt · knowledge basen8n on Railwayone workflow per booking systemBooking systemPhorest · Barespace · Floomly · Fresha · Salon IQ
Messaging
Messenger · WhatsApp · InstagramDM agentsame tools · same rulesn8n on Railwayshared orchestration
Underneath
Supabase PostgresRLS · edge functionsClient & agency dashboardsBetter Stack · Slack digestsalerts · morning briefs

— Shipped for

CallHandlers.ioAI receptionist platform · Ireland & UKSalons & Beauty12–15 live businesses on voice + DM agentsHeating & Energy SupplyOrders & deliveries verticalSyndrixProject-based software & automationInfosoft StudioMini Clean Business Solutions · DavaoRegional GovernmentResources management moduleAgribusinessMulti-farm inventory · TacurongRetailMulti-branch POS platform

02Experience

From shipping code
to running the system.

Three roles, one direction: more ownership of what happens after deploy. Software engineer first; now technical operations and AI systems for a live fleet.

Aug 2026 — Present
CallHandlers.io
Remote · Ireland / Germany team
Production ownership · Voice & messaging AI
Technical Operations Manager · AI Systems Engineer
Sole engineer on production for an AI receptionist platform serving 12–15 live salons and service businesses across Ireland and the UK. I own the stack from the voice layer down: agent behaviour, orchestration, database, integrations, infrastructure and monitoring.
- Own agent behaviour across the fleet: system prompts, knowledge bases, tool contracts and voice/ASR settings, tuned from live call transcripts, with business rules enforced in workflow code rather than prose.
- Design and operate the orchestration layer: multi-tenant n8n on Railway, five booking-system integrations (Phorest, Barespace, Floomly, Fresha, Salon IQ, several reverse-engineered) and Supabase Postgres behind row-level security.
- Built the omnichannel DM agent for Facebook Messenger, WhatsApp and Instagram and integrated it with the phone receptionist, so both run on the same tools, rules and client data, with automatic hand-off the moment the owner types. Live, still being tuned from real conversations.
- Plan the platform’s architecture decisions: where each rule lives (agent prompt, workflow code or database), how tenants stay isolated on the shared chassis, and how new channels and verticals plug in without forking the stack.
- Optimize and reconcile client data across 12–15 live tenants, so the agents, the knowledge bases and Postgres all answer from the same facts.
- Run production end-to-end: uptime monitoring, morning digests to every client channel, incident diagnosis, zero-downtime changes with rollback plans, and the findings register the team works from.
- Ship the operator tooling (client and agency dashboards, live callback queue, number pool, credential registry) and the platform’s first non-salon vertical for a residential heating & energy supplier, with UI optimization across the site and dashboards.
- ElevenLabs
- n8n
- Railway
- Supabase
- PostgreSQL
- TypeScript
- Twilio
- Better Stack
- Slack
May 2024 — Present
Syndrix
Project-based · Philippines
Contract · Full-stack & automation
Software Engineer | AI Automation Specialist
Contract full-stack and automation work for small businesses: I gather the requirements, design the database, build the app, deploy it and stay on after launch.
- Built a multi-branch POS and inventory platform (React, Node.js, Express, PostgreSQL, Prisma, TypeScript) running real-time sales and stock replenishment across three retail locations.
- Designed and deployed a farm-to-warehouse inventory system (Laravel, Inertia.js, Vue, MySQL) across three farms and two warehouses for 9–17 daily users, with role-separated dashboards and zero data loss.
- Automated business workflows with n8n, Zapier and the Gemini API — from an AI recruitment pipeline to HubSpot and Slack syncs — and run CI/CD on GitHub Actions with Docker and Nginx on DigitalOcean at 99.9% uptime.
- Use Claude Code, Cursor and MCP tooling daily, and verify the output with tests — syntax is cheap, judgement is not.
- React
- Node.js
- Express
- Laravel
- Vue
- PostgreSQL
- MySQL
- Prisma
- n8n
- Gemini API
- Docker
- DigitalOcean
Sep 2025 — Jan 2026
Infosoft Studio
Mini Clean Business Solutions · Davao City
486-hour OJT under a Chief Software Architect
Software Engineering Intern
Backend-focused internship on a regional government platform — from REST endpoints and access control to a large legacy migration.
- Shipped 15+ Laravel REST endpoints for a regional government resources module now in production, with a granular RBAC model for multi-office hierarchies and bulk permission updates.
- Automated the migration of 4,500+ government resolutions and 2,000+ PDF documents with a custom Python pipeline at 99%+ record integrity, on schedule.
- Built an Express.js backend for a case-records system with a standardized REST response contract adopted by the frontend team.
- Mentored five fellow interns through the Laravel request lifecycle and MVC, and represented engineering in client stakeholder meetings.
- Laravel
- PHP
- MySQL
- Express.js
- Python
- Docker
- Laradock
- Ubuntu

03Selected Work

Systems that run
in production.

The flagship is the platform I operate today. The rest are products and client systems I designed and shipped end-to-end — three with full case studies.

01/ 04
2026 — PresentCallHandlers.io · Production
Live · 12–15 clientsCase Study Live Site
AI Receptionist & DM Agent
Read case study
Voice and messaging AI that books the real diary
Production AI receptionist for salons and service businesses across Ireland and the UK. Each client gets an ElevenLabs voice agent with its own compiled prompt and knowledge base; every booking action runs through a shared n8n chassis on Railway and one of five booking-system integrations. The DM agent brings the same receptionist to Facebook Messenger, WhatsApp and Instagram.
CallerTwilioElevenLabsn8n · RailwayBooking systemSupabase
- Owns agent behaviour across the fleet: prompts, knowledge bases, tool contracts, voice settings
- DM agent on Messenger, WhatsApp & Instagram sharing the receptionist’s tools and rules
- Zero-downtime Postgres migration US-West → EU: 12× faster queries, sub-second replies
- First observability stack: Better Stack, status feeds, watchdogs, morning digests
- Ghost-probe replay harness proves every fix against the live booking system
- 129 production findings documented with severity, root cause and proof
- ElevenLabs
- n8n
- Railway
- Supabase
- PostgreSQL
- TypeScript
- Twilio
- Better Stack
02/ 04
2025 — PresentIndependent · Product
Independent projectCase Study GitHub
ScanServe
Read case study
Real-time multi-tenant queue management
A real-time, multi-tenant queue platform that streamlines customer flow across organizations, branches and service windows. QR-code registration issues live tickets to phones, WebSocket boards replace manual refreshes, and a five-tier hierarchy scales from one office to a multi-branch institution.
- Selected for presentation at regional PSITS and Expo events for institutional and government adoption
- QR-code registration issuing live tickets (e.g. REG-001) with mobile status
- WebSocket live display boards for call / serve / no-show events
- Hierarchical roles: Super Admin · Org · Branch · Service Point · Window
- Laravel
- Livewire
- Alpine.js
- Tailwind CSS
- MySQL
- WebSockets
- Docker
03/ 04
2025 — PresentIndependent · Product
Independent projectCase Study GitHub
EggFlow
Read case study
Farm-to-market production & distribution ERP
End-to-end management system tracking the complete lifecycle from farm collection to final sale across multiple locations. Shift management with clock-in tracking and shift-end discrepancy calculations enforces cash and stock accountability; executive dashboards replace weekly manual reports.
- Lifecycle tracking from farm collection to multi-location sale
- Shift discrepancy automation enforcing cash and stock accountability
- Executive dashboards: production comparisons, sales trends, inventory valuation
- Redis caching and SendGrid transactional notifications
- Laravel
- Livewire
- Alpine.js
- PostgreSQL
- Redis
- SendGrid
- Docker
04/ 04
2026Syndrix · Automation
Delivered
AI Recruitment Pipeline
Automated candidate screening with n8n and Gemini
An automated screening workflow that ingests PDF applications, extracts structured qualifications with the Gemini API, scores candidates against the role, syncs records to HubSpot and posts audit-ready summaries to Slack — replacing manual first-pass review.
- PDF applications parsed into structured qualifications with the Gemini API
- Candidate scores and summaries posted to Slack for human review
- HubSpot records created and kept in sync automatically
- Every decision traceable: input, extracted fields, score, reviewer
- n8n
- Gemini API
- HubSpot
- Slack
- Webhooks

03.1More work

01
Corporate Egg Production System
Syndrix client · inventory & delivery across 3 farms and 2 warehouses
Laravel · Inertia · Vue · MySQL
Live with client
02
Retail POS & Service-Log Platform
Syndrix client · stock, returns and repairs across 3 branches
React · Express · PostgreSQL · Prisma
Live with client
03
Regional Government Resources Module
Infosoft Studio · 15+ Laravel endpoints, RBAC, 4,500+ records migrated
Laravel · MySQL · Python
In production
04
Case Records Management System
Infosoft Studio · Express.js backend with standardized REST contracts
Express · Node.js · Docker
Delivered
05
UECFI Church Hymnal App
Independent · offline-first hymnal with chord transposition
React Native · Expo · SQLite · Firebase
GitHub ↗

04Capabilities

What I build,
and keep running.

Four kinds of engagement — voice and messaging agents, orchestration and integrations, platform and reliability, and full-stack product work. Each one owned end-to-end, from scoping to the on-call after launch.

01— Capability
Voice & Messaging AI Agents
Conversational agents that do real work: answer the phone or the DM, check a live diary, and book, move or cancel — with behaviour you can predict.
- ElevenLabs conversational agents: prompt architecture, knowledge bases, voice & ASR settings
- Tool-calling contracts and evaluation against real call transcripts
- Latency tuning and model selection across Claude, Gemini and GPT
- DM agents for Messenger, WhatsApp and Instagram with owner hand-off
02— Capability
Orchestration & Integrations
The layer where the real work happens: multi-tenant n8n workflows, webhooks, and integrations with systems that were never meant to have an API.
- Self-hosted n8n on Railway, one shared chassis per integration
- REST API reverse-engineering from browser traffic to a working adapter
- HubSpot, Slack, Twilio, Zapier and webhook-driven pipelines
- Health checks for credentials and sessions that expire silently
03— Capability
Platform, Data & Reliability
Production ownership: the database, the infrastructure, the monitoring, and the runbook you reach for at two in the morning.
- Supabase / PostgreSQL: row-level security, edge functions, migrations, egress tuning
- Zero-downtime migrations with rollback cards and abort deadlines
- Observability with Better Stack, status feeds, watchdogs and daily digests
- Railway, DigitalOcean, Docker, Nginx, Cloudflare Tunnels, GitHub Actions
04— Capability
Full-Stack Product Engineering
Dashboards, consoles and business systems people use every day — built end-to-end from requirements to deployment.
- React, Next.js, TypeScript and Tailwind on the front
- Laravel, Livewire, Express and Prisma on the back
- Multi-tenant architecture, RBAC and real-time features
- Client onboarding flows and internal tooling for operators
- UI optimization across public sites and operator dashboards

05Method

How I operate,
written down.

Four habits from running live systems for real businesses — each one learned from a specific incident, and each one leaving an artefact you can check.

01phase
Diagnose
Read the artefacts before forming an opinion — call transcripts, database rows, workflow logs, vendor behaviour. Every finding gets a severity, a root cause and a named proof, or it doesn’t go in the register.
Artefacts
- Findings register entry
- Reproduction or probe
- Severity & blast radius
02phase
Design
Decide where a behaviour belongs before writing it: rules in workflow code, facts in retrieval, per-client values in config. Prompts drift; node code doesn’t. Write the rollback before the change.
Artefacts
- Layer decision
- Rollback card with abort deadline
- Escalation note when it’s a founder call
03phase
Ship
Small, reversible, off-hours when it touches live clients. One table at a time, verify after each. Replay real cases through a harness before and after — it has caught bugs in my own fixes.
Artefacts
- Before / after probe results
- Live agents verified
- Commit with the evidence
04phase
Operate
Ownership continues after deploy: uptime monitoring, deploy-freshness watchdogs, a morning digest per client, a 72-hour watch on every launch, and runbooks so the next person doesn’t repeat the incident.
Artefacts
- Status channel & alerts
- Daily digest
- Runbook & lessons entry

06About

An engineer who
operates what he ships.

Background, education and the toolbelt — from the first Laravel endpoint to running a fleet of voice agents.

— Bio

I build and run AI systems in production. Right now that is voice and messaging agents for 12–15 businesses: ElevenLabs agents, n8n orchestration on Railway, multi-tenant Supabase Postgres, five booking-system integrations, and the monitoring and incident work that keeps them up. Before that I shipped full-stack Laravel and Express systems for government, agribusiness and retail clients.

Name: Mark Anthony Resoso
Role: AI Systems Engineer · Technical Operations
Company: CallHandlers.io
Location: Tacurong, Philippines
Languages: Filipino · English
Timezone: UTC+08:00

06.1Toolbelt

Voice & Conversational AI

ElevenLabs Conversational AI
System prompt architecture
Knowledge bases (RAG)
Tool-calling contracts
Voice & ASR tuning
Latency optimization
Transcript evaluation

LLMs & Agents

Claude Code
Gemini API
GPT
Model evaluation & selection
Model Context Protocol (MCP)
Cursor
Agent skills

Orchestration & Integrations

n8n (self-hosted, multi-tenant)
Webhooks
Twilio
Messenger · WhatsApp · Instagram DMs
HubSpot
Slack
Zapier
API reverse-engineering

Data

PostgreSQL
Supabase (RLS, edge functions)
MySQL
Redis
SQLite
Schema design
Query tuning

Infrastructure & Reliability

Railway
DigitalOcean
Docker
Nginx
Cloudflare Tunnels
GitHub Actions
Better Stack
Sentry
Zero-downtime migrations
Rollback runbooks

Languages

TypeScript
JavaScript
Python
PHP
SQL
Bash
PowerShell

Frontend & Product

React
Next.js
Vue
Inertia.js
Livewire
Tailwind CSS
React Native (Expo)
TanStack Query
UI optimization

Practice

Technical operations
Architecture decisions
Incident diagnosis
Root-cause analysis
Findings registers
Client onboarding
Technical writing for non-technical stakeholders

06.2Proof points

01
Moved a production database across continents with zero downtime — 12× faster queries, sub-second caller replies, 88/88 verified runs, rollback card written before touching anything.
02
Closed the crash loop behind five production outages, fixed at 02:06 local while the salons were shut, with every live agent verified afterwards.
03
Root-caused an accent-drift defect that predated my hire to contradictory audio tags between the voice config and the system prompt — and presented it to the founder as a decision, not a unilateral fix.
04
Documented 129 production findings in an institutional register; a replay harness caught a bug in my own fix before it shipped.
05
Designed and shipped two independent products end-to-end (ScanServe, EggFlow); ScanServe was selected for regional PSITS and Expo presentation.
06
President’s Lister and Dean’s Lister, BSIT 2026 — plus a Gemini-powered study companion built in two hours at the GDG Davao hackathon.

07Contact

Have a system
that needs running?

I'm open to AI systems, backend and platform roles: remote, contract, or a team that needs someone to own production. Email is fastest; I reply within 24 hours.