Julie Bosworth
Software Developer
Full-stack software developer. I build reliable software that works for the people using it. Currently at AFMEPF and pursuing an MS in Computer Science at Georgia Tech.
Experience
American Federation of Musicians and Employers' Pension FundNov 2024 – Present
Software Developer
- •Build and maintain full-stack features for MAX, an enterprise pension platform serving 30,000+ union musicians.
- •Develop staff-facing tools in Visual Basic, including an address-conflict detector that flags when an employer's address already appears on another record.
- •Develop and maintain SQL stored procedures, tables, and triggers for MAX's pension management workflows.
- •Build and schedule JAMS jobs for recurring workflows, including a monthly employer payment reminder that attaches agreement documents and logs categorized delivery failures.
- •Built an engagement payment reminder job estimated to save staff roughly a week of manual email processing per month.
- •Updated the benefits application to capture applicant type for dual-eligible participants.
- •Built a direct deposit self-service portal in Visual Basic, Java, and SQL, covering routing number validation, image upload, SSRS PDF generation, and automated email and work request management.
FAST EnterprisesSep 2020 – Jul 2024
Software Implementation Consultant
- •Implemented features for a business tax system that collects over $1B a year for the NYC Department of Finance, working directly with client stakeholders to gather requirements.
- •Wrote SQL queries surfacing business metrics from the NYC DOF tax database, producing reports consumed directly by Finance stakeholders.
- •Tested and maintained features across collection and tax-return processing workflows.
- •Replaced a manual multi-hour review with a nightly job that detected missing tax files, recovered and processed them, and emailed staff a status report.
Projects
PokepeekLive
Jan 2026- •Built a full-stack Pokémon search platform in Next.js and TypeScript that visualizes PokéAPI data with interactive charts.
- •Wrote Playwright end-to-end tests covering desktop and mobile flows, and deployed the site on Vercel.
- •Shipped a stat-distribution explorer and a type-matchup analysis page, using Claude Code to iterate on the UI.
Next.jsTypeScriptPlaywrightPokéAPIVercel
Next RecLive
May 2026- •Built a media recommendation engine with a Next.js/TypeScript front end and Python/FastAPI backend, using OpenAI embeddings to rank recommendations by semantic similarity.
- •Applied scikit-learn's k-means clustering to group similar media into themes and auto-labeled each with GPT-5 mini.
- •Built a labeled test set of known cross-domain matches and used it to compare embedding inputs, raising top-1 recommendation accuracy from 35% to 55%.
- •Built an interactive WebGL map of the catalog's embedding space, iterating on the UI with Claude Code.
- •Migrated the backend from GCP Cloud Run to Vercel.
Next.jsTypeScriptPythonFastAPIOpenAIscikit-learnGCPVercel
Skills
Languages
PythonJavaJavaScriptTypeScriptVisual BasicSQLHTMLCSS
Frameworks & Libraries
ReactNext.jsFastAPIFlaskscikit-learn
Databases
PostgreSQLSQL Server
Tools & Platforms
GitDockerAWSGCPVercelPlaywrightOpenAI APIClaude Code
Education
Georgia Institute of TechnologySep 2025 – Present
M.S. in Computer Science
University of Massachusetts AmherstMay 2020
B.S. in Computer Science & Psychology