David Umanzor
I'm a Developer
About
Hello! My name is David, I am an aspiring developer who is passionate about AI, Machine Learning and backend work. I enjoy working on projects in my free time between testing full stack applications and open source contributions.
Data Scientist & Software Developer
- Birthday: 30 Sept 2000
- Phone: +863 529 7685
- City: Orlando, USA
- Age: 23
- Degree: Bachelors
- Email: deumanzor0@gmail.com
- Freelance: Available
Resume
DownloadSummary
David Umanzor
Driven and aspiring developer with 1+ years of experience in AI and Machine Learning, as well as multiple experiences with research.
- Orlando, FL
- (863) 529-7685
- deumanzor0@gmail.com
Education
Bachelor of Science & Computer Science
2022 - 2025
University of Central Florida, Orlando, FL
Associate of Arts & General Arts
2019 - 2021
Polk State College, Lakeland, FL
Organizations
Theta Tau
Jan 2022 - Present
University of Central Florida, Orlando, FL
KnightHacks
Jan 2022 - Present
University of Central Florida, Orlando, FL
Google Developers Student Clubs
Jan 2024 - Present
University of Central Florida, Orlando, FL
Phi Theta Kappa Honor Society
2019 - 2021
Polk State College, Lakeland, FL
Vice President of Leadership
- Collaborated with other officers to develop and execute Honors in Action (HIA)
- Hosted the 2021 Annual Florida Regional Conference for Phi Theta Kappa, facilitating engaging workshops and guest speakers
- Contributed to the development of the college’s quality enhancement plan to maintain accreditation
- Awarded 2022 All-Florida Academic Team
STEM Club
Aug 2020 - Dec 2021
Polk State College, Lakeland, FL
President
- Designed and facilitated personalized, progressional team building exercises to foster group development
- Became an LSAMP Scholar by completing leadership, global studies, and academic research requirements
- Mentored members to increase engagement in classes and success towards degree
Professional Experience
Machine Learning Engineer
May 2023 - Aug 2024
CAE USA Inc, Orlando, FL
- Developed data processing pipelines: Created a repository for processing column-based data and applying machine learning models using Sklearn, XGBoost, and LightGBM. Utilized Kubeflow and Docker to automate hyperparameter tuning with Ray Tune, optimizing model selection and performance.
- Collaborated on behavior trees for simulation agents: Designed and implemented a collection of 9 core behaviors and multiple subtrees using Rust and XML for autonomous agents within Computer Generated Forces (CGFs) simulations. Contributed to white papers detailing the architecture and documentation for intelligent agents.
- Implemented MinIO, Trino, and Iceberg on Kubernetes using Helm charts, laying the groundwork for a scalable data analytics layer to process simulation data from CGFs. This work set the stage for future expansion, including UI integration and enhanced data accessibility, as part of a collaborative effort within a 3-person team.
Undergraduate Researcher
Jun 2021 - Dec 2021
Florida Agricultural & Mechanical University
- SMS Spam Detection with Transformers
- Employed a Natural Language Processing model called DistillBERT to analyze a data set of spam and non-spam text
- Conducted using python libraries of sklearn, numpy, and pandas to preprocess data and train the model
- Achieved a high level of accuracy, with a Receiver Operating Characteristic (ROC) Curve of 98% True Positive Rate
- Identifying ASL with Machine Learning and AI
- Employed Convolutional Neural Network model called Keras to accurately identify ASL signs
- Conducted research using python libraries of numpy, cv2, seaborn, and sklearn to preprocess data and train the model
- Presented research at the Florida Undergraduate Regional Conference (FURC) 2022
Experience
Mathematics TRiO Tutor
Jun 2021 - Dec 2021
Polk State College
- Aided students in completing homework assignments, identifying lagging skills, and correcting weaknesses
- Tutored students one-on-one and in small groups in high school and college-level math, resulting in improved grades
- Demonstrated expertise in mathematics concepts, teaching techniques, and study skills to foster greater student engagement
Portfolio
- All
- Full Stack
- Tool
- AI/ML
Campus Connect
Developed a web application to connect students to local events around their university.
Fintracker
A Web application designed to help users track their finances and budget.
Spam SMS Detection using Transformers
Training an Hugging Face DistilBERT model to detect spam SMS.
Identifying ASL using Convolutional Neural Networks
A program to train an model using CNN methods to accurately detect the letter in ASL.