Senior Machine Learning Engineer

Michaels Stores, 04/2022 - 01/2025

• Carried out proof-of-concept initiatives to enhance search systems, recommendation engines, and content moderation.
• Researched, designed, and developed quantitative methodologies, machine learning models, algorithms, and tools.
• Developed, enhanced, built, and tested predictive models.
• Designed and developed machine learning and deep learning models (including linear regression, logistic regression, decision trees, random forests, clustering, CNNs, and transformers) to improve text classification, sentiment analysis, visual search, and product tagging.
• Performed ad hoc quantitative analysis and conducted exploratory data analysis (EDA) to identify patterns, trends, and anomalies for model development and feature engineering.
• Optimized models using statistical and machine-learning techniques for visual search and perception systems.
• Investigated and applied state-of-the-art computer vision methods for representation learning and visual feature extraction.
• Utilized transformer-based large language models (LLMs) such as BERT, RoBERTa, DeBERTa, SBERT, MiniLM, FLAN-T5, GPT-2, and GPT-4 to enhance search results.
• Created and maintained technical documentation for modeling, including project plans, model descriptions, mathematical derivations, data analyses, processes, and quality controls.
• Conducted model performance analysis, quantified model limitations, provided comprehensive interpretations, explanations, and conclusions.
• Delivered technical presentations and reports to both technical and non-technical audiences, explaining model performance and key findings.

Research Associate

University of Colorado Colorado Springs, 02/2020 - 04/2022

• Formalized science of artificial intelligence for open-world novelty.
• Investigated novel approaches integrating statistical extreme value theory to enhance machine learning models.
• Innovated algorithmic solutions to elevate accuracy in computer vision triplet tasks.
• Devised unsupervised novelty characterization and adaptation for open-world deep learning classifiers.
• Trained and fine-tuned a range of deep learning models, including Convolutional Neural Networks (CNNs) and Transformers.
• Boosted performance of open-world image classifiers by 18%.
• Improved the mean accuracy of change-points detection by 50% using statistical extreme value theory.