Jameson Railey

• Member of the company’s first team to build a production application entirely with agentic AI (Claude Code), accelerating delivery by 38% and moving the release from October 2026 to July 2026.
• Architected an LLM-powered AI assistant chatbot for 1,000+ FreeFlow Vision Connect users, owning it from prototype to production with an agentic development workflow.
• Co-authored an engineering prompt library of 20 phase prompts and 16 agent skills — designing prompts and agent logic for an agentic software development lifecycle — and facilitated 5 company-wide AI ideation sessions reaching 100+ engineers.
• Developing a Next.js and TypeScript canvas application hosted on Azure that lets users apply spot color, foil, and varnish embellishments to PDFs across a variety of print materials; shipped a beta release to 200 users after 7 months of development.
• Built a TypeScript and Next.js dashboard for the business team, visualizing 10+ engagement metrics across 1,000 users of FreeFlow Vision Connect.
• Modularized Azure resource deployment using DevOps, CI/CD and Bicep, increasing deployment success by 50%.

• Built Python data-processing logic and deployed it as serverless Azure Functions, ingesting 1,000+ daily printer job records from 100+ devices into Azure Table Storage.
• Integrated the Azure Functions pipeline with the FreeFlow Vision Connect production web app to surface previously unavailable usage data such as sheet counts, media types and other job metrics in near real time.
• Designed Power BI dashboards on top of the new dataset that gave the business team visibility into how FreeFlow Vision Connect customers were using their devices, informing customer-engagement and product decisions.

• Solo-building a map-first fitness event discovery platform for iOS, Android, and web, sharing types, auth, and UI across a TypeScript monorepo of Expo React Native and Next.js apps styled with Tailwind CSS.
• Owning the full lifecycle from architecture through implementation with an agentic AI workflow in Claude Code, modeling the Supabase Postgres schema and validating every payload with Zod against generated TypeScript types.
• Enforcing strict TypeScript and ESLint module boundaries across the workspace, testing with Jest and Playwright, and gating every pull request on lint, type checks, and a production build through GitHub Actions.

• Co-developed a React-based social media platform for image sharing and editing with a team of 4, using Fabric.js for in-app image editing and Firebase Auth for user registration and session management.
• Implemented GraphQL client queries against the Apollo Server backend for image retrieval and user-profile fetching, with assets stored in MongoDB and cached via Redis.

• Began as an independent study with a Bucknell professor: a Scorigami is an NFL game ending in a score that has never happened before, and the question was whether the next one could be predicted.
• Built Monte Carlo and time-based Poisson simulations in Python (pandas, scipy) that model NFL game outcomes by sampling scoring events from empirical and Poisson-fitted distributions over 1M simulated games per scenario.
• Showed that a plain Poisson model misses real game dynamics, then extended it to a non-stationary, state-dependent Poisson process that fits actual score data more closely and forecasts likely future Scorigamis.
• Second of four authors on the resulting paper, published in the Proceedings of the 2024 Winter Simulation Conference.

Read the paper (opens in new tab)

Stevens Institute of Technology

Master of Science · Men’s Soccer Team

Bucknell University

Bachelor of Science in Business Administration · GPA 3.83/4.0 · Men’s Soccer Team