Alirio Hernando Martínez Barreto

Specializing in optimizing large language models (LLMs) for accuracy, reliability, and real-world applicability. - Evaluated and enhanced models deployed by industry leaders (e.g., OpenAI, Google, Meta), ensuring outputs meet strict standards for factual correctness, logic, and bias mitigation. - Improved model performance by diagnosing errors in reasoning, code, and data workflows, delivering actionable feedback to engineering teams. - Automated Python-based QA pipelines for data science projects, reducing manual review time and accelerating model iteration cycles.

Selected for an intensive, project-based engineering program, delivering end-to-end AI solutions for real-world scenarios in e-commerce and fintech. Focused on building production-ready models and automating data workflows under industry-standard practices. - Image Classification API: Built and deployed an API for product image classification achieving 91% accuracy. Containerized with Docker for consistent deployment and performed load testing to ensure stable performance. - Credit Risk Model: Developed a scoring model to predict loan defaults using customer financial data. Applied feature engineering and model validation techniques, achieving reliable predictions on unseen data with minimal overfitting. - Automated Data Pipelines: Created end-to-end pipelines that automatically process data from multiple sources (APIs, CSV files), perform cleaning and feature engineering, enabling faster model training cycles. - Technical Skills Applied: Python, Scikit-learn, Pandas, Docker, API development, data preprocessing, model evaluation (ROC AUC, accuracy metrics), and exploratory data analysis.