Stephen V. Brown

Stephen V. Brown

NLP Researcher

Founder of LyraText

With over a decade of experience as a full-time software engineer and a background in computer science, I've always enjoyed building solutions for complex problems. I founded LyraText in 2024 to combine my NLP research with my software expertise to help other researchers and businesses solve the difficult challenges found in analyzing narrative data.
I was previously on the business school faculty of the accounting departments at Arizona State University and the University of Connecticut, along with a visiting position at the University of Florida. My research focuses on what we can learn from language in corporate disclosures beyond what’s obvious to a human reader and how information about one company can tell us more about other companies.

Research Interests

Corporate Disclosures
Natural Language Processing
Computational Linquistics
Artificial Intelligence
Deep Learning

Education

Ph.D. in Business Administration
University of Florida
Master of Business Administration
University of North Florida
B.S. in Computer Science
University of North Florida

Publications

Expertise

I have around 20 years’ experience as a software engineer, designing and building a wide variety of object-oriented libraries, databases, end-user applications, and back-end systems. Most of my software development work now is in direct support of other researchers through my company, LyraText.

Programming Languages

Python Especially Polars, spaCy, PyTorch, Pandas, and scikit-learn; nearly all of my machine learning/AI work is in Python
SQL Extensive database development experience; in recent years pretty equally divided between SQL and noSQL backends, often PostgreSQL and MongoDB
R I use the tidyverse where possible; most of my R development is for ad hoc statistical analysis and data visualization
JavaScript Primarily to support web frontends; preferred libraries/toolsets are Svelte, SvelteKit, and Tailwind CSS

General Data Science

Analytical Approach Deep learning, machine learning; traditional statistics and econometrics
Data Transformation Processing, transforming, and merging large, semi-structured data; full extract-transform-load (ETL) process; I generally prefer Polars to Pandas for dataframe work
Data Types Accounting, financial, operational, and natural language
Statistical Tools Primarily R for traditional statistics; Python for deep/machine learning; Stata occasionally
Data Visualization ggplot2 in R; matplotlib in Python

Development Environment

Operating System I have extensive Linux server administration experience and prefer Linux for both server and desktop use; I regularly use Fedora, Ubuntu, and Pop!_OS
IDE Neovim for everything possible

Contact