Full-stack developer & designer
ElizabethStein.
I design and build software that’s actually running in production. Eighty-six things shipped — here are the eight that matter, live right now.
The problem
Managed-service providers had no fast, credible way to measure a client’s cybersecurity readiness — assessments lived in spreadsheets and consultants’ heads.
What I built
A guided assessment on Dynamics 365 / Power Platform with a Next.js front end. Twelve phases from data model to scoring to reporting, shipped solo across the full stack.
Proof
Live in production and in use by a cybersecurity nonprofit. Sole developer — data model, business logic, front end, and deploys.
Context
Agency work — a portfolio of client sites that each need to ship fast, stay maintainable, and share a design language without becoming copy-paste.
What I do
Build and maintain eleven Craft CMS sites in PHP and Twig, with reusable components and content models tuned for non-technical editors.
The idea
Enter any date from 1958 to 2020 and get the cultural snapshot — the #1 song, the movies, the prices, the news — synthesized into an era-aware narrative. Not “what was #1 in 1975” but what that day felt like.
How it works
A date is parsed, then four Algolia indices (Billboard, TMDB, FRED prices, Wikimedia events) are queried in parallel and an LLM synthesizes the results into one narrative, rendered in a shareable CRT/VHS aesthetic. Next.js 16 with Upstash rate-limiting and Langfuse LLM observability.
Why it matters
The one project here with an externally-verified outcome: it won the Algolia Agent Studio Challenge against a judged field of entries.
What it is
Point it at a repository and it narrates how the code fits together, exposing 14 Model-Context-Protocol tools so an AI agent can explore the codebase with structure instead of guesswork.
The idea
Most screenshot-to-code tools hallucinate components that don’t exist in your system. Trace grounds generation against a real catalogue, so what comes out is code you can actually merge.
What it is
Feed it your sources and it answers documentation questions with RAG — a full product with subscription tiers, auth, and Stripe billing wired in, not a demo.
Why it exists
The most lived-in thing here: the command-line tool I built to run my own operation — one interface over ~90 repositories and 10 services. I use it every day, which is the real test.
What it is
Practise interviews out loud with a voice agent on the OpenAI Realtime API, with FSRS-5 spaced repetition scheduling what you review next. A live SaaS.