Alberto Rosales Email & Phone Number

Santa Clarita, CA, US

• LLM Performance Optimization: Researched and implemented advanced techniques to evaluate and refine the final output generated by different Large Language Models, achieving high quality results to clients.
• Debugging and Code Fixing: Worked with AI Large Language Models by guiding their thought process to generate trustworthy code fixing bugs and syntax errors, perfecting model understanding and output.
• Industry-Wide Expertise: Usage of deep technological expertise to assist the team workflow to be able to support major sector-wide organizations, including some of the biggest companies.
• AI LLMs Advanced Training: Played an important and pivotal role in the implementation of Reinforcement Learning with Human Feedback to assist clients building robust AI models, improving critical business insights, and.

• Developed a classification model for hospitalization prediction of elderly population, achieved a 94% ROC/AUC score using XGBoost and K-Folds, built an API with a form for user prediction.
• Created an NLP model for sentiment analysis of movie reviews, processed non-traditional data, trained custom word embedding with Word2Vec, achieved an 89% ROC/AUC score in opinion detection.
• Innovated a CNN model for vehicle classification, trained MLP, LeNet, and ResNet50 models on 196 classes, attained an 83% accuracy with ResNet50.
• Implemented a ML API system, receiving images for content prediction using ResNet50, accomplished a 93% prediction score, utilized Redis, containerized with Docker, and conducted stress tests with Locust.
• Created a regression model for credit risk scoring with an 84% ROC/AUC score using visualization libraries, feature selection, outlier handling, normalization, standardization, and categorical feature encoding.
• Built an ELT data pipeline for e-commerce that reduced the time to analyse customer data by 50% using SQL queries and analysing the relationship between holidays and purchase order dates.

Guadalajara, Jalisco, Mexico

• Coded a data-driven Covid-19 web scraper for Mexico, extracting and aggregating case/death counts, and news updates. Established automated email updates for users, scored 9.5/10 in project evaluation.
• Streamlined Excel processes using UiPath Studio's Framework, creating reusable components for enhanced efficiency, scored 8.8/10 in project evaluation, showcasing potential for data handling and analysis.
• Earned several UiPath diplomas, completing 50 courses in business analysis, process mining, robotic process automation, and UiPath tools, accumulated a total of 640 hours of intensive training, solidifying proficiency.

Colima, Mexico

• Developed, updated, and optimized API features using programming languages and tools, enabling the frontend team to meet sprint tasks on time; accelerated application development progress from 50% to 70% in record time.
• Implemented data-driven insights to improve cross-team collaboration, resulting in up to 40% increase in overall workflow efficiency.
• Led Freeway training program for backend members using data-driven evaluation methods, onboarding 3 additional fellows, enhancing the overall team capabilities and expertise.