Data Analyst and Machine Learning Engineer. I work on NLP and multimodal model projects applied to education.
I am Gary Vidal, Data Analyst and Machine Learning Engineer. I currently work on AI projects applied to education, designing and evaluating tasks with multimodal models and natural language processing techniques.
I have also developed time-series predictive models, data pipelines, and optimization solutions for different contexts. You can contact me here.
Check my music project! I'm the guitarist and backup singer of the progressive metal/rock band Automata.
I also produced, mastered, mixed and engineered the band's first EP "Memento Mori" in 2025.
AI Analyst (Education and NLP)
May 2025 - Present
CIAE
Design and evaluation of educational tasks with multimodal models, response-quality analysis, and development of text extraction pipelines from images.
Data & ML Intern
Jan 2023 - Feb 2023
Logaritmo Information Technologies
Development of data pipelines and time-series predictive models (ARIMA), plus solutions for call center workforce planning.
Mining Optimization Intern
Jan 2022 - Feb 2022
Alicanto Labs
Implementation of optimization models with dynamic programming, MILP, and Gurobi for mining problems.
MSc in Applied Mathematics
2024 - 2025
Universidad de Chile
Thesis area: Local certification,
certification algorithms.
Thesis title:
Local certification of graph classes defined by forbidden
cycles: minor case and induced case.
Thesis advisors: Ivan Rapaport, Pedro
Montealegre.
Mathematical Engineering
2017 - 2025
Universidad de Chile
Awards: Outstanding student 2017 and 2018.
General Engineering, Double Degree Program
2019 - 2021
Centrale Marseille Engineering School
Desktop overlay for Football Manager 2024 that imports squad exports and ranks players by role, position, and formation.
View details
Development of Bayesian classification models focused on CART trees and Markov-chain-based optimization for dataset alignment.
View details
FastAPI service that predicts music genre from WAV audio files, including model metadata, readiness checks, and inference endpoints.
View details
