Dylan Bates Email & Phone Number

-Collaborated with both Engineering and Research teams to develop and build out air-gapped generative AI models, including EdgeRunner Tactical: the top performing 7B parameter LLM in the world.-Spearheaded advanced training and optimization of Low-Rank Adapters (LoRAs) for diffusion models (SDXL, FLUX) and low VRAM solutions, enhancing model efficiency across edge devices.-Pioneered integration of diffusion models and transcription/translation models on Intel NPU, driving cutting-edge research on low-resource edge AI deployment.

-Participated in The Erdős Institute's Spring and Fall 2022 Data Science Boot Camps, covering Python, machine learning, and quintessential data science techniques-Served as a five-time Teaching Assistant, guiding other fellows through a dozen synchronous problem solving sessions covering material from lectureSpring 2022: -Lead a team to complete a top 5 final project using player and team-level data to predict who will win the 2022 NHL Stanley Cup Championships-Improved test accuracy beyond the theoretical upper bound of 62%Fall 2022: -Lead a team to complete a second top 5 final project using convolutional neural networks to predict where wildfires will spread-Improved the area under the precision/recall curve to nearly 30% (best ever for this dataset)Summer 2023:-Helped lead the inaugural Deep Learning Boot Camp, covering topics from Recommendation Systems to Attention-based TransformersSpring 2024:-Lead a team to complete top 5 final project in the Deep Learning course, using traditional and modern Machine Learning techniques (XGBoost, CNNs, audio-transformers) to predict music genre-Achieved 92% accuracy on GTZAN (1000 tracks, balanced), and 63% F1 score on FMA Medium (25,000 tracks, unbalanced)-Served as a project judge, reviewing 12 Data Science Boot Camp teams' final projects for Technique, Creativity, Presentation, and Overall Results.Summer/Fall 2024:-Served as a project judge again, reviewing Data Science Boot Camp teams' final projects for Technique, Creativity, Presentation, and Overall Results.

Hartsville, South Carolina, United States

-Director of the Center for Quantitative Literacy, Coker's math tutoring center-Math Program Coordinator and Course Coordinator for MAT-103 (Quantitative Literacy), MAT-203 (Intro to Statistics), MAT-220 (History of Math), MAT-230 (Precalculus), MAT-232 (Calculus II), and MAT-250 (Mathematical Reasoning)-Mentored students with mathematical modeling projects, and served as faculty critique. -Taught a full double load (24 credit hours) during Fall 2021.

Raleigh, North Carolina, United States

As the Graduate Services Assistant for the Center for Research in Scientific Computation.I ran the CRSC research lab on campus, maintaining the equipment, assisting faculty with collecting experimental research data, and assisting students with mathematical modeling, alongside my own research in the lab.I mentored three students as part of the Global Training Initiative GEARS: providing Global Education, Academics, and Research Skills to talented undergraduate students, who performed original research using virtual reinforcement learning to balance a real inverted pendulum.I helped four undergraduate students collect data and run experiments as part of Directed Research for Undergraduates in Math and Statistics (DRUMS), culminating in an award-winning poster presentation "Using Modeling and Machine Learning to Characterize POTS" at the NCSU Summer Undergraduate Research Symposium.

Raleigh, North Carolina, United States

I taught a variety of live and virtual classes, including Math 242 (Calculus III), and Math 131/231 (Calculus for Life and Management Sciences A/B). Responsibilities include up to 8 hours of lecture per week, creation of quizzes, homework assignments, and tests; and office hours.

Raleigh, North Carolina, United States

I graded, reviewed material, administered quizzes, and prepared students for homework and tests. I also lead recitation for Math 242 and Math 131, which involved reviewing material with smaller groups of students, and going over practice problems to prepare for exams.

Columbia, South Carolina, United States

I taught Math 142 (Calculus II), Math 115 (Precalculus) and Math 111 (Basic College Mathematics). Responsibilities include 4-6 hours of lecture per week, creation and grading of quizzes, homework assignments, and tests; and office hours. I also worked as a Teaching Assistant from 2015-2016 for Math 142 (Calculus II), which included Maple labs, reviewing material, administering quizzes, and preparing students for homework and tests.

Hartsville, South Carolina, United States

Once a week, I taught a computer science or robotics course at the local high school through PULSE, which is designed to improve academic achievement and personal development of students. I oversaw tests, administered and provided assistance with labs, and ensured that students understand the workings of the programming language.

Raleigh, North Carolina, United States

Working with a team of graduate and undergraduate researchers at the Developer Liberation Front, I helped create a social screencasting system. Performing original research, I wrote a paper entitled Recommending More Efficient Workflows to Software Developers, which was submitted to and accepted at the SPLASH Student Research Competition in Portland, Oregon. I presented this research in October 2014, earning 2nd Place for the undergraduate division.

Hartsville, South Carolina, United States

Summer 2012: Seeking to improve placement outcomes, I used data mining to determine if there was a correlation between high school GPA and success in various math classes. Later, I used item response theory to determine which questions best differentiate between students of low and high abilities to come up with a better placement test. An unsurprising byproduct of this effort was an increased number of students enrolled in remedial classes.Summer 2013: Building off of research conducted in the summer of 2012, I created an online transition course for placing students into MAT-101 who had failed the mathematics placement test, which was modified the year before. The final product included lectures, four tests, hundreds of questions, and links to videos explaining the concepts students were struggling to understand. By recording all student activity on the placement exam and transition course, the faculty was able to analyze trends and adapt instruction and preparation quickly to best serve incoming students.The results of this research were presented at the Celebration of Academics Undergraduate Research Symposium in 2013 and the Scholarship of Teaching and Learning Commons Conference in 2014.

Doctor Of Philosophy - Phd, Applied Mathematics

Thesis: Virtual Reinforcement Learning for Balancing an Inverted Pendulum in Real Time Advisor: Dr. Hien Tran

Master Of Arts - Ma, Mathematics

Activities and Societies: Preparing Future Faculty Credential with DistinctionThesis: Subdivision of Measures of Squares Advisor: Dr..

Bachelor Of Arts - Ba, Mathematics And Computer Science, 4.0

Activities and Societies: Dean's Commendations, Cross Country Captain, Computer Science and Mathematics Club - President, Astronomy Club.