Will Smitas

• Relevant coursework: Data Structures and Algorithms, Statistics & Probability, Partial Differential Equations, Abstract Algebra, Machine Learning.

• Relevant coursework: Complex Analysis, Basic Analysis, Linear Algebra, Multivariable Calculus.

• Won first place among all competitors at an in-person competition at Jane Street's New York City office.
• Competition involved solving complex mathematical problems, along with the design and development of trading strategies.

Software Engineer Intern

• Developing custom AI tooling using Python and the Ollama API for 20+ small businesses and their marketing teams.
• Presented an analysis to business owners on prompting techniques and benchmarks for OpenAI and Anthropic LLMs.

Research Assistant

• Audited Lean 4/Mathlib formalizations against natural-language mathematical statements for a large-scale autoformalization dataset, verifying logical faithfulness across combinatorics, algebra, analysis, and number theory.
• Identified subtle semantic gaps between natural mathematical language and formal code, including missing domain restrictions, weakened existential quantifiers, and unstated type generalizations.
• Performed clause-by-clause verification requiring fluency across many advanced subfields of mathematics to determine faithfulness of formal statements to natural language.

Undergraduate Research Assistant · Albany, NY

• Investigated a potential link between group representation theory and Diophantine analysis.
• Designed Python programs to compute 50,000+ integer solutions for multivariable polynomials and analyze the properties of and patterns within those solutions.
• Created a poster summarizing the research and presented it about 20 times to UAlbany community members.

Brown University

• Led development of automated Kalshi trading, written in Go and Python, deployed on AWS EC2.
• Trained machine learning models on Brown University's supercomputer, Oscar.
• Developed a framework in Go for distributed market data to parallelize market trade streams, divided over web sockets and HTTP polling.
• Scraped millisecond-latency crypto data from Coinbase, Kalshi, and Polymarket.

Quantitative Trading at Brown

• Built a Monte Carlo simulation engine in Python (NumPy) modeling Geometric Brownian Motion to generate thousands of future stock price paths.
• Derived an empirical CDF from simulated outcomes to estimate the probability of a stock falling below a given price threshold.
• Validated simulation accuracy by benchmarking results against the Black–Scholes closed-form model.
• Visualized price path distributions and CDF outputs using Matplotlib.

Brown University Athletics

• Balance a rigorous academic schedule with a 20-hour weekly commitment to Division I athletic training and competition.