Provo, Utah, United States
I am a Master’s student in Computer Science at BYU with published research experience in swarm intelligence and multi-agent systems. I have advanced programming skills in Python and experience with decision-making algorithms, reinforcement learning, knowledge graphs, modeling, and simulation. Committed to sustainability and interested in applications of AI in Autonomous Systems and Robotics for building a better, safer, and more efficient world. ---------------------------- Personal Interests: Outside of computer science, I enjoy teaching Lindy Hop and other styles of historic dance. I danced on the BYU Folk Dance team for two years and competed with the Salt Shakers, a Lindy Hop team based in Salt Lake City. I love music, rhythm, and learning new hobbies.
• Grading assignments for 20+ students, evaluating understandings of network properties, centrality measures, modularity, community detection, knowledge graphs, and machine learning methods; and providing written feedback. • Identifying common areas of confusion and providing actionable feedback to the instructor to improve assignment clarity and course learning outcomes. Material Covered: • Network Topology: Random Graphs, Small-World (Watts-Strogatz), Scale-Free (Barabási-Albert). • Centrality Measures: Eigenvector, Katz, Betweenness, PageRank, Degree Centrality. • Community Detection: Louvain, Girvan-Newman, Newman Hill-Climbing, Laplacian/Spectral Graph Cut • Machine Learning & Data Science: Node2Vec, Knowledge Graphs, Graph Convolutional Neural Networks (GCNs) • Tools: NetworkX, Gephi
Contributed to research on the use of Grammatical Evolution for automatic generation of cooperative behaviors in multi-agent systems. Grammatical Evolution Framework: Systems Architecture: • Engineered a modular Discrete-Time Agent-Based Modeling (ABM) simulation engine and framework for Grammatical Evolution; Utilized a Self-Evaluating Abstract Syntax Tree (AST) architecture with stack-based program execution to decouple experiment design from grammar parsing and interpretation logic. • Integrated a transactional rollback system to detect cycles and revert invalid state changes during crossover and node mutation, thus maintaining the structural integrity of resulting programs. • Created a Domain-Specific Language (DSL) for grammar definition using Python context managers, meta classes, and decorators; allows grammars that are context-aware and self-validating, preventing the creation of invalid programs. • Eliminated boilerplate code and streamlined the end-to-end pipeline for world/agent definition, simulation setup, and visualization, reducing new experiment setup time by weeks. Performance: • Utilized efficient sparse-matrix encoding, state-change tracking, and on-the-fly reconstruction for creating memory-efficient animations; only records when toggled, to reduce computational load across hundreds of simulations. • Implemented a high-efficiency subtree crossover algorithm that performs structural recombination while maintaining the integrity of the Abstract Syntax Tree (AST). Utilizes auto-updating indexed node lists to achieve constant-time O(1) lookup for individual node compatibility testing. Then uses systematic set-based elimination to achieve O(n) lookup for overall program compatibility checking. • Achieved a 10x speedup in program generation by engineering a decorator-based class factory to migrate validated and static instance attributes to the class level, reducing redundant initialization overhead across millions of nodes.
• Multi-Agent Simulation: Implemented several models of swarm behavior, including Reynolds’ Boids (vector-based steering) and Couzin’s Three-Zone Model, to serve as benchmarks and to gain familiarity with Swarm Dynamics. • Stochastic Modeling: Independently modeled a multi-agent Best-of-N problem as an Absorbing Markov Chain. Extracted transition probabilities (for the collective state of agents) using Monte Carlo methods, to predict convergence to optimal or sub-optimal sites. • State-Space Analysis: Leveraged Strongly Connected Component (SCC) Analysis to evaluate predictive limits of the state representation; found ~80% of states produced convergence probabilities in a narrow band around the global baseline (~20% optimal), indicating minimal predictive signal. • Theoretical Analysis & Embeddings: Proved the functional equivalence of Graph Convolutional Neural Networks (GCNs) to a binomial expansion on the transition matrix in certain cases; validated their use as dimensionality reduction tools, collapsing high-dimensional collective states into low-dimensional embeddings, while still capturing the Sufficient Statistics of the underlying Markov process. • Publication: Co-author of Puneet Jain et al. ‘Performance Prediction of Hub-Based Swarms,’ Philosophical Transactions of the Royal Society A (2025). [DOI: 10.1098/rsta.2024.0141]
• Technical Feedback: Graded assignments for 20+ students, evaluating understandings of network properties, centrality measures, modularity, community detection, knowledge graphs, and machine learning methods; and providing written feedback. • Student Mentorship: Conducted 1-on-1 sessions to clarify complex mathematical and algorithmic concepts, supporting student learning and understanding. • Instructional Support: Facilitated technical review sessions and led discussions to expound on core course concepts and help students prepare for exams. • Content Engineering: Refined and enhanced existing project specifications and instructions to improve clarity and reduce implementation ambiguity for complex graph-based programming tasks. Material Covered: • Network Topology: Random Graphs, Small-World (Watts-Strogatz), Scale-Free (Barabási-Albert). • Centrality Measures: Eigenvector, Katz, Betweenness, PageRank, Degree Centrality. • Community Detection: Louvain, Girvan-Newman, Newman Hill-Climbing, Laplacian/Spectral Graph Cut • Machine Learning & Data Science: Node2Vec, Knowledge Graphs, Graph Convolutional Neural Networks (GCNs) • Tools: NetworkX, Gephi
• Technical Initiative: Independently mastered the tools to deploy a suite of in-house applications and automated workflows for Dining Administration and Retail Restaurants, with zero prior experience in most of the stack (Wrike, Microsoft Power Platform, Azure); Leveraged existing expertise in Python, Excel, SQL, and Databases to deliver production-ready solutions. • Systems Integration: Built automation pipelines using Wrike REST APIs, Power Automate, Azure Functions, and Excel (VBA) to streamline reporting, task assignment, and support ticket workflows; reduced ticket response times and improved cross-team communication across all of Dining Services. • Performance Optimization: Resolved a critical reporting bottleneck by migrating a legacy triple-nested for loop in Microsoft Power Automate to Azure Functions. Used Python and Pandas DataFrames to reduce total processing time from 120 minutes to 2 minutes (98.3% reduction). • Full-Stack Development (Agile): Built a Pizza Order Fulfillment and Delivery application using SQL Server, Bite Order API, and Microsoft Power Apps to bypass current ERP limitations; delivered ~90% of core functionality within a two-week sprint, anticipating stakeholder needs in advance to accelerate deployment and reduce iteration cycles.