General Virtual Assistant

Leo Mari Cuizon · QA · General Virtual Assistance · AI projects

Hire me View Upwork Profile

Mobile apps · Web · General Virtual Assistance · Basic AI workflows and Creative work

Portfolio overview screenshot

About

I work across QA, AI tools, automation, research, and practical day to day operations.

I started with data entry, online research, administrative tasks, and simple Canva work before moving into Software QA and system validation. More recently, I've been exploring AI-assisted workflows, automation, and lightweight products through personal projects.

Today, I work across Software QA, AI tools, and creative projects. I use AI throughout my process for research, troubleshooting, writing, development, and workflow automation, while continuing to build small projects to learn new tools through hands-on work.

Location
Cebu City, Philippines
Focus
QA · Startup Operations · Virtual Assistance · AI
Availability
Open to work · Remote
Approach
Practical thinking · Structured execution

What I Do

What I actually do.

01

Mobile & Web App QA Testing

  • Funtinal, exploratory, and smoke testing
  • User flows, auth/session bugs, forms, and UI behavior
  • Reproducing and documenting bugs with clear steps and evidence
  • Validating fixes and checking for regressions
02

AI, Automation & Tool Support

  • Using ChatGPT, Claude, Codex, n8n, Make, and other AI/no-code tools
  • Building and testing simple workflows and lightweight tools
  • AI assisted research, troubleshooting, documentation, and development.
03

Virtual Assistance & Research

  • Online research and lead generation
  • Data entry, spreadsheets, and documentation
  • Basic WordPress and Creative work
  • Following briefs, checklists, and existing processes
↳

I also build some of the tools I work with

  • Built a personal PWA notes app and a simple automated workflow for a repetitive task
  • Building these projects gives me hands-on experience with both the user side and the underlying workflow
  • These projects helped me in terms of learning new tools and exprimenting with practical ways to improve repetitive tasks

How I Work With AI

AI, QA, virtual assistance, and creative work with a focus on structured and reliable execution.

I use AI and other tools to support practical work across QA, research, administration, automation, and creative projects. I focus on understanding the task, following the required process, and checking the output before considering it done.

→

QA & Testing

  • Exploratory, smoke, and functional testing
  • Bug reproduction, documentation, and validation
  • Reviewing outputs for accuracy and consistency
→

Virtual Assistance & Operations

  • Data entry, spreadsheets, and documentation
  • Lead generation and administrative support
  • Following briefs, SOPs, structured processes, and learning unfamiliar tools.
→

AI & Automation

  • Using ChatGPT, Claude, Codex, Make, and other AI/no-code tools
  • Using AI for research, writing, troubleshooting, and development
  • Connecting tools through APIs and webhooks when needed
→

Creative Work

  • Short-form video editing and basic motion/content work
  • Canva graphics and simple marketing assets
  • Research, scripting, generation, editing, and QA of AI-generated content

Projects

Things I built to understand
how systems break.

Personal experiments — not polished products. Each one was a reason to get closer to a real failure mode.

Job Intel MVP screenshot
Personal experiment · v47+

Job Intel MVP

Rule-based job evaluation workflow that scores remote listings against a candidate profile — without AI ranking. Transparent, rule-based logic produces explainable outputs. Built a multi-GPT workflow to manage the build: a source-of-truth Hub, a specialist GPT for architecture decisions, and Codex for narrow implementation tasks with explicit constraints on what each tool could decide.

Visit project
Stackr screenshot
Personal experiment · Three.js

Stackr

Offline-first AI notes PWA iterated across 47+ versions. I used my own apps to observe and debug real failure cases during development. Caught a critical auth failure caused by iOS Safari's Intelligent Tracking Prevention blocking Supabase session persistence on PWA reinstall — isolated the caching conflict and documented the fix path.

Visit project
Jungle Dash screenshot
Personal experiment · PWA

Jungle Dash

2.5D endless runner PWA built as a testing ground for continuous-state systems: collision detection, mobile control behavior, obstacle generation edge cases, state resets on game death, and performance under sustained loops. Object pooling, garbage collection pressure, and service worker behavior under offline conditions — all testable in a way most apps don't expose.

Visit project

Live Demo

AI Bug Triage Automation

Paste any raw bug report. GPT-4o-mini classifies severity, scores priority, and extracts structured fields via a Make.com automation pipeline.

Live
Bug report input 0 chars
Try a sample:

What You Get

Every engagement ends with
something you can act on.

Documented outputs, not activity summaries. Here's what that looks like in practice.

📋

Bug Reports

  • Steps to reproduce, exact conditions, expected vs actual behavior
  • Severity classification and suggested fix path
  • Delivered in your preferred format — doc, sheet, or Notion
🧪

Lightweight Web, PWA & Automation Builds

  • Simple HTML websites, lightweight PWAs, landing pages, and AI-assisted product experiments
  • Mobile responsiveness, offline behavior, auth/session handling, and workflow validation
  • Small automation prototypes using tools like n8n, Make, APIs, and AI-assisted workflow logic
🎬

AI Content Workflow Production

  • Short-form content workflows documented from brief to export
  • AI voice, avatar, captions, editing, and tool constraint notes
  • Best for lightweight UGC ads, workflow demos, and portfolio-style creative samples
🗂️

Dataset Curation

  • Structured text datasets extracted and cleaned for LLM training pipelines
  • Consistent labeling, formatting, and deduplication across large collections
  • Source-verified, schema-consistent, delivered in your required format

Sample Bug Report

Real report, sanitized client details
P1 — Blocker

System fails to generate downstream outputs after successful data processing

Context

Web application · Staging environment · Workflow: Data ingestion → Processing → Output generation

Steps to Reproduce

  1. Create a new workspace/entity
  2. Connect a data source and initiate processing
  3. Allow processing stage to complete successfully
  4. Trigger output generation step A, then step B
  5. Observe output status

Expected

Both output generation steps complete, producing valid output artifacts.

Actual

Processing completes. Both output steps fail. No artifacts created.

System Logs (Sanitized)

processing completed successfully (items_processed: 16, blocked: false)
output_a.asset_id = null
output_b.asset_id = null
last_failed_runs.output_a.status = failed
last_failed_runs.output_b.status = failed

Analysis (LLM-assisted QA review)

Processing completes successfully, but output generation fails. This suggests a possible issue in how processed data is passed into the output generation step, or missing required input mapping between stages.

LLM Evaluation

Real evaluation, sanitized client details

Multi-model document summarization — 6 models, 4 document lengths

I conducted structured evaluation of multiple LLM outputs across different document lengths, assessing accuracy, reasoning, completeness, and hallucination behavior. Claude Opus was used as the primary reference model for consistency and comparison.

Rubric (6 dimensions)

Grounding · Reasoning · Completeness · Actionability · Clarity · Task Overlay — plus hallucination flags: none / minor / critical

Key findings

  • Gemini 3.1 Flash Lite fabricated phantom citation brackets [1][2][3] throughout a 40K-word task — classified hallucination_critical, scored 33/100
  • GPT-4.1 nano fabricated a numerical statistic by an order of magnitude in a medium document — hard-to-catch and high-risk in production
  • Haiku 4.5 fabricated an institutional affiliation in a very long document — unacceptable in clinical or high-stakes contexts
  • GPT-4.1 mini was the top short and very long document summarizer; Sonnet 4.6 was most reliable for medium and long

Output — routing recommendation

Task Primary Fallback
Short (~1K words) GPT-4.1 mini Haiku 4.5
Medium (~5K words) Sonnet 4.6 Haiku 4.5
Long (~15K words) Sonnet 4.6 Gemini 2.5 Flash
Very long (~40K words) GPT-4.1 mini Sonnet 4.6

Mini Case Study

AI-assisted debugging & documentation

PWA login sessions disappearing on refresh

Problem

Users were logged out every time the PWA was refreshed or reinstalled on iOS Safari.

Tested

Service worker caching strategy, Supabase auth token storage, ITP cookie behavior across iOS versions.

Output

Surfaced a caching conflict blocking session persistence. Documented the issue and fix path with AI assistance.

Contact

Have a project that needs an extra pair of hands?

I can help with QA and testing, virtual assistance, AI tools and automation, documentation, or creative projects. I'm comfortable learning new tools and working from a brief or existing processes. Remote and open to freelance, part-time, and project-based work.

Contact:

Upwork Email LinkedIn