Jorge Mora

HR and recruiting teams were tracking candidates, roles and pipelines across disconnected spreadsheets, slow, error-prone and impossible to audit.

An end-to-end talent-management platform with a single source of truth: candidate pipelines, role tracking, and role-based access, designed to replace the spreadsheet sprawl.

Centralized hiring operations into one auditable system, with a data model built to grow alongside the team.

Optical retail needed a way to detect and analyze eyeglasses on a face in real time, without sending video to the cloud or relying on a constant connection.

An on-device computer-vision pipeline using MediaPipe for face landmark detection and a lightweight ONNX model (~6.2 MB), with WASM and WebGPU acceleration support.

Real-time detection running fully on-device, keeping inference private and responsive without server round-trips.

Users had no single place to check all phone lines registered under their CURP, making it impossible to detect identity misuse or unauthorized registrations.

A fast web app that aggregates mobile carriers and MVNOs in Mexico, letting users check all registered lines via a single CURP query in real time.

Gave users a quick, reliable tool to detect unauthorized phone registrations and identity misuse.

NASA has one of the largest image archives ever assembled, but its native search tool is too literal to surface what people are actually looking for.

A better interface over NASA's public API, fast, clean, and unaffiliated, that makes the archive genuinely explorable.

Turned a sprawling public archive into a browsable, searchable interface that feels designed for humans.

Editorial and business teams ran critical workflows by hand: scheduling shifts, tracking payroll, commissions, loans, with no shared state and frequent mistakes.

An internal platform modeling newsroom operations and administrative processes: journalist scheduling, payroll, commissions, expenses, and employee loans in one system.

Centralized day-to-day operations, eliminated manual spreadsheet errors, and gave management a reliable audit trail.

A radio news station had no unified system to manage employee activities, automate Facebook publishing, and oversee newsroom operations.

A business web app integrating WordPress and Facebook APIs to manage the radio station, supervise employee activities, and automate social media processes.

Replaced manual processes with an integrated system that reduced coordination overhead and automated content publishing.

Identifying songs from live microphone audio requires low-latency inference with high accuracy across a large fingerprint database.

A real-time music detection system using convolutional neural networks to identify songs from live audio input through the user's microphone.

Built a working Shazam-style system demonstrating real-time CNN-based audio fingerprinting in a web environment.

American Sign Language recognition from camera input requires fast, accurate ML inference running in a web context with real-time feedback.

A real-time sign language recognition web application using machine learning to detect and classify ASL alphabet letters through camera input.

Delivered a working real-time ASL classifier accessible entirely in-browser without any native app installation.

DICIS students had no fast way to find available classrooms or professors in real time, relying on outdated printed schedules.

A lightweight tool that web-scrapes the university schedule system and surfaces available classrooms and professors in real time.

Gave students an instant, mobile-friendly way to find free spaces on campus without talking to administration.

The thermodynamics lab at the University of Guanajuato had no web presence for students to access resources, schedules, or lab information.

A web application built for the thermodynamics lab at DICIS, providing students and staff access to lab resources and information.

Delivered a modern, maintained web presence for the lab used by students and faculty at the Division of Engineering.

Producing weather forecast videos with a consistent format required time-consuming manual editing across assets, scripts, and narration.

A desktop app that automates forecast video production using FFmpeg for rendering, Whisper for speech-to-text, and NLP to process and structure weather data.

Cut video production time significantly by replacing manual editing with a semi-automated pipeline with consistent output quality.

A grocery store needed a fast, reliable way to weigh products and print labels with weight and barcode without a complex POS system.

A lightweight desktop application that connects to a scale, captures weight, and prints formatted barcode labels for grocery products.

Gave the store a simple, dependable tool that reduced checkout errors and replaced a manual labeling process.