Bruno Gutiérrez
Mexico City · CDMX

Forward Deployed AI Engineer

I build AI systems that work
in production — and prove they do.

Real-time voice, multi-agent pipelines, and the evaluation work that tells you whether a change was an improvement or just a change. Everything below is mine, and it’s all running — scroll and see it, no clicking required.

See it runningGet in touch
Bruno Gutiérrez
PhDComputer Science · MSc Mathematics
Ex-WalmartData Scientist
Sub-secondreal-time voice, in production
4 systemsdesigned, built and shipped solo

Selected work

Live nowAlex, the AI English teacherAlex is live right now.▶ Talk to Alex — try it now

One minute, in your browser. No sign-up, no install — headphones and a mic, in Chrome.

English Circle

Live

An AI English-conversation tutor for Mexican Spanish speakers

Live voice classes with “Alex,” an AI teacher you actually talk to. Alex replies in natural spoken English in well under a second, understands Mexican slang, corrects you gently without breaking your flow, and sends a personal report after every class.

~500 msspoken response time
2voice architectures benchmarked
600+automated tests
PythonLiveKit AgentsGemini LiveDeepgram STTElevenLabs TTSNext.js
How I built it →
“Preciosa Mentira” — teaser, playing silently. Produced in a single day, start to finish. Watch it with sound →
Luna — a character generated by the pipeline
Luna · character
El Chopo — a location generated by the pipeline
El Chopo · location

Cinema Pipeline

In progress

An AI pipeline that produces cinematic CDMX shorts

An AI pipeline that produces a complete cinematic world set in Mexico City — original characters, locations, storyboards and short films. Press play: the film, the faces and the streets all came out of the pipeline. No stock, no actors, no camera.

100%generated — no stock, no actors
One worldcharacters, locations and film
PythonGoogle ADKGemini 2.5Imagen 4Generative videoMulti-agent pipeline
How I built it →

Real-Guitar Music Game

In progress

Play Mexican music on a real guitar

Plug a real guitar into your computer and play along. It listens to what you actually play — detecting your chords in real time — scrolls a rhythm lane, shows finger positions, and the crowd reacts as you nail the changes.

Real timechord detection from live audio
60 FPSgame loop
PythonPygameReal-time audio DSPChord detectionSQLite
How I built it →

Research

PhD Research — The Moving Firefighter Problem

Optimization & deep reinforcement learning for fire containment

Published

My doctoral work: a novel graph-theory model for fire spread and containment, solved two ways — exact mathematical optimization (MIQCP) and deep reinforcement learning — then compared head to head. Published, peer-reviewed.

PythonDeep Reinforcement LearningMathematical OptimizationMIQCPCellular Automata

Looking for a forward-deployed role

Sitting close to the problem, scoping it with the people who own it, building the system, and being accountable for whether it moved the number.

Get in touchLinkedIn