Portfolio

Tomas Ibañez

Bilingual operations assistant who builds with AI

I help teams stay organized and move faster. I handle customer communication in English and Spanish, keep tasks and processes in order, and use AI tools to build simple websites, workflows and documentation.

tomasag.ibanez@gmail.comBuenos Aires, Argentina (GMT-3)Available full-time: remote, hybrid or on-site

Claude & Claude CodeNetlifyGitHubMicrosoft OfficeGoogle WorkspaceClickUpZapierAI image generationEnglish & Spanish

What I bring

Three things I do well

Customer service

Calm, friendly service in busy, high-pressure settings, from VIP guests at South America's largest stadium to families at one of Buenos Aires' largest malls.

Organization

Calendars, inboxes, tasks and processes. Trained as an executive assistant, and I document how work gets done so anyone can repeat it.

Building with AI

I use AI tools every day to research, write and build: websites, interactive tools, email sequences and step-by-step guides.

Featured project · Live & open source

VOD Clipper: a desktop app that turns long streams into ready-to-post clips

Context: personal project, built for a content creator friend Timeline: 4 days, Nov 2026 Role: idea, spec, directing Claude Code, testing, publishing
VOD Clipper: hours of stream in, the best minutes out

The problem

A friend edits his stream recordings with Claude, but a 1–3 hour video is several GB. Claude.ai accepts 500 MB per file and doesn't watch video, so he was cutting every stream into 12 parts by hand.

The idea

Don't send Claude the video. The app turns it into a small pack Claude can read (transcript, thumbnails, energy signals). Claude picks the moments, and the app cuts them from the original file at full quality.

What I built

  • A Windows app with a guided 5-step workflow, in Spanish and English
  • Three outputs: one highlight video, separate clips, or vertical Shorts with captions
  • Feedback on every result, and an app that learns how you like to edit
  • A landing page, an open-source GitHub repo and a build guide (SOP)
6 GB → 15 MBA 3-hour stream becomes a pack Claude can read
10 clipsFrom a 17-minute conversation, ready for TikTok
10–20 sTo auto-edit each vertical Short
200+ testsAutomated checks on every part of the app

The value for creators

  • Hours of cutting become minutes: pick a recording, choose what you want, and get clips, a full highlight video or Shorts.
  • No quality loss: the final files are cut from the original recording.
  • Private and free: everything runs on the creator's own computer, and no video is ever uploaded.
  • Gets better with use: ratings and trims teach the app each creator's style.

How I used AI

  • Planning: wrote a one-page spec with Claude: the problem, the idea, the limits and the first version.
  • Building: directed Claude Code with one instruction file per feature, each with a clear "done when" list, plus a progress log so long overnight runs could resume where they stopped.
  • Reviewing: had Claude review the finished app for privacy, security and licensing before publishing.

What testing caught

Sound

In the highlight video, the audio drifted a little further out of sync at every join. A test measured it and the fix brought every join back in sync.

Speed

Transcribing on a PC without a graphics card was slow. I measured it, tuned the settings and added faster options.

Security

The review found that any open website could talk to the app's local server. It's now locked so only the app can.

Bootcamp final project

A client acquisition system, built with AI

Context: Pareto Talent Executive Assistant Bootcamp, final project Timeline: 5 days, Oct 2026 Role: planning, building, writing, testing

The goal

A remote-staffing company wanted more qualified business owners booking sales calls. That meant a free resource owners would actually want, and a way to send only good-fit leads to the sales team.

What I built

  • A landing page, built with Claude Code and published on Netlify
  • An interactive 10-minute diagnostic tool
  • A short qualification survey that routes each lead
  • Automated follow-up emails in a CRM
  • 5 ads with AI-generated images
  • A launch guide (SOP) and a project board in ClickUp

Result

  • A live system, tested end to end on every path
  • Every lead gets something useful; only good fits are invited to book
  • Earned the bootcamp certificate

How it works

1 · ATTRACT5 adsOne for each type of customer
2 · CAPTURELanding pageBuilt with AI, live on Netlify
3 · QUALIFY4-question surveySize, budget, timing, needs
Good fit
Booking pageThe diagnostic tool, then a calendar to book a call
Follow-up emailsReminders until they book, then a confirmation
Not a fit yet
Thank-you pageThe same diagnostic tool, no rejection
Helpful emailsFree resources, door left open

How I used AI

  • Planning: researched the target customer and competitors, and shaped the offer with Claude.
  • Building: created the landing page and the diagnostic tool with Claude Code, then published them on Netlify.
  • Writing: drafted page copy, the 5 ads and the email sequence, then edited them by hand.
  • Visuals: generated the ad images with AI image tools.

What testing caught

Logic

One rule used AND instead of OR, so every lead was marked as a good fit. I found it in the logs and fixed it.

Tools

The form could only send people to one page, so I rebuilt it as a survey with routing rules.

Shared account

In an account shared by 160+ users, other people's automations fired on my test contacts. I labeled and filtered everything so my system only touched my own leads.

Proof

See it for yourself

Experience

Where I've worked

Jun 2026 – Aug 2026

Host & Cashier · Neverland, Abasto Shopping

Family entertainment center in one of Buenos Aires' largest malls. Welcomed and guided visitors in peak season, handled cash and card payments, kept stock organized.

May 2025 – Nov 2025

VIP Area Service & Operations · Estadio Monumental (River Plate)

The largest stadium in Argentina and South America. Served high-profile guests and kept the VIP area running smoothly during major events.