Sean R. B. Bearden

Data Scientist and Researcher

Sean Bearden Ph.D.

 

Portfolio Highlights

 
 

Resume Chatbot: Harnessing Chatgpt for the job search

Highlighted Skills: Python, LangChain, LLM (ChatGPT), AWS SAM, DynamoDB, Heroku, Dashboards

Presentation: Understanding my PhD research - The Search For King Turing’s Stolen Treasure

Highlighted Skills: Keynote, Non-Technical Presentation Style, Graphic Design

EFFICIENT SOLUTION OF BOOLEAN SATISFIABILITY PROBLEMS WITH DIGITAL MEMCOMPUTING

Highlighted Skills: MATLAB, Dynamical System Design, Technical Writing, Graphic Design

Screen Shot 2020-05-06 at 6.58.10 PM.png
 

About

some background info

Sean Bearden recently earned his Ph. D. in Physics as a member of Dr. Massimiliano Di Ventra’s group at the University of California San Diego. He researched the application and development of memcomputing systems, a novel computing paradigm considered to be a classical alternative to quantum computing. Memcomputing is an efficient computing paradigm that uses memory to process and store information in same physical location. While many constraint satisfaction problems are approached by employing algorithmic solvers, their group leverages memory dynamics by designing nonlinear dynamical systems for particular problems, such as Boolean satisfiability and combinatorial optimization. The memcomputing technique can be applied to many real-world problems, such as, scheduling, searching, and optimizing (see case studies).

Shortly after Sean joined the research group, it was decided by MemComputing, Inc., a startup partially owned by Dr. Di Ventra, to part ways from the academic research group at UCSD due to patent concerns. Sean was given the difficult task of creating a unique memcomputing system that was independent of the implementation used at the company. After much frustration, failure, and assistance from group members, Sean was successful at creating a new system that could efficiently solve Boolean satisfiability problems that were previously difficult for his research group to solve. 

Identifying as a nontraditional learner, Sean was a bit of an underachiever prior to college. Sean found his love of education later in life by realizing he could learn on his own terms, utilizing his strengths and finding alternatives to his weaknesses. It was a process he had to explore on his own, but Sean later drew upon these introspective years to tutor other students whose learning styles were underrepresented in the classroom. He brings his unique approach to research, often resulting in unexplored possibilities, aiding the process of discovery.

Furthermore, Sean identifies as a lifelong learner. At any point in time, he is usually engaged in learning a skill outside of his PhD research. Past pursuits include accelerated MBA programs, programming language certification, and leadership and teamwork training. Beyond acquiring skills, Sean enjoys using his skills for creative projects. These can range from Arduino and Raspberry Pi projects to creating algorithms for daily fantasy sports prediction. 

These creative projects are especially alluring to Sean when a project requires merging several skills to create something intricate. For example, Sean’s mother once sent him a Deadpool mask as a gift for Halloween. The Deadpool had been released earlier that year, so Sean thought many people might dress up as Deadpool. However, he did not want the mask to go to waste, thinking to himself, “Wouldn’t it be cool if the Deadpool head played soundbites from the movie? I wonder if I can make it happen…” So, Sean figured out how to play audio files from an Arduino microcontroller connected to small speakers. To control which soundbite plays, he connected a button pad and programmed the Arduino to play sixteen Deadpool audio files. Once programmed, Sean carved out the mannequin head to fit the electronics inside, adding light up eyes for additional effect. The final result can be seen here.

Growing up in Buffalo, New York inspired Sean to move very far away from snowy winters, but he is still a connoisseur of chicken wings. To alleviate the stress that is inevitably coupled with graduate research, he enjoys training Brazilian Jiu-Jitsu at the P5 Academy in San Diego.

Programming Languages

Relative usage of programming languages in my research.
 

Interests

Data Science
Machine Learning
Large Language Models
Public Speaking
Financial Markets

Screen Shot 2020-05-06 at 6.58.10 PM.png

Experience

Valmar Holdings L.L.C., Data Scientist and Chief Analytics Officer

November 2020 - October 2023

  • Designed and managed unified underwriting system using integrated scikit-learns decision trees, catering to diverse lending portfolios summing to millions of dollars funded monthly.

  • Developed and implemented a lead purchasing volume forecast model to meet funding goals across various lending portfolios, by integrating call center performance metrics, attendance data, anticipated funding from alternative sources, and refinancing projections.

  • Led the evaluation of underwriting products and developed prototype solutions for fraud detection.

  • Collaborated with Morgan Stanley on the investment management of a multi-million-dollar portfolio, overseeing a range of financial activities and executing risk management strategies.

  • Reduced lead purchasing cost-per-funding to lowest levels in company history.

  • Modified defaulted loan write-off policy to reduce collection efforts on customers by 50% without significant reduction to funds collected prior to write-off.

UC San Diego, Massimiliano Di Ventra Group, Graduate Student Researcher

September 2016 - October 2020

  • Designed and developed a dynamical system that can solve optimization problems with one million variables and millions of constraints.

  • Compared successful optimization algorithms against the MemComputing paradigm, generating and collecting more than 10,000 problem instances for meaningful comparison.

  • Modified dynamical system to sample the mode of the model distribution of a restricted Boltzmann machine by solving a QUBO problem mapped into a MAX-2-SAT problem.

  • Studied recurrent neural networks and used an echo state network for time-series forecasting of cellular automata.

Wyzant, Private Tutor

June 2015 - September 2019

  • In-person and online tutor for over 100 high school and college students in various physics and mathematics courses, catering to students with alternative learning styles.

  • Consulted for futures trader, generating results counter to contracted developer, saving more than $10,000.

SUNY at Buffalo, Igor ZutiC Group, Student Researcher

January 2013 - July 2015

  • Simulated spin-laser operation in Mathematica, generating data to analyze and showing optimal digital operation occurs with unconventional material properties.

  • Swept parameter space of coupled nonlinear dynamical system to discover non-monotonic trends that assisted the development of ultrafast spin-lasers.

Research Areas

Relative focus of my academic research.
 
Screen Shot 2020-05-06 at 6.58.10 PM.png

 Publications

Screen Shot 2020-05-06 at 6.58.10 PM.png
Screen Shot 2020-05-06 at 6.58.10 PM.png

Education

UC San Diego

Ph.D. PHYSICS
2020

UC San Diego

M.S. PHYSICS
2018

SUNY at Buffalo

B.S. PHYSICS AND APPLIED MATHEMATICS
2015

 

Ohio University, Athens

A.S NATURAL SCIENCES
2011

Awards

UC President’s Dissertation Year Fellowship
National science foundation graduate research fellowship
Alfred P. Sloan University center for exemplary mentoring Fellowship
UCSD SUmmer Research Fellowship
UCSD Physics Excellence Award
Barry Goldwater Scholarship

Contact

Email: seanbearden@seanbearden.com

Screen Shot 2020-05-06 at 6.58.10 PM.png