Full-stack engineer · Provo, Utah

Full-stack engineer building dependable software.

I work across TypeScript, AI evaluation, and developer tooling—especially where reliability matters and a quiet failure has real consequences.

My AI math tutor, taai, is live in invite-only alpha. Request access and try it out.

Nelson Schnepf

Currently building from Utah.

Things I’ve shipped.

Software is most interesting when it earns a place in someone’s daily life. These are tools I have built, released, and kept useful beyond a demo.

01

Stint

An open-source macOS menubar time tracker for ADHD-friendly work, with a paid tier secured by Ed25519 license verification.

ElectronReactTypeScriptSQLite
02

sp CLI

A TypeScript terminal client for the open-source Super Productivity desktop app, spanning 11 command groups and an extended local API.

TypeScriptCommanderZodVitest
03

Focus Coding

The production monorepo behind a tutoring business: a Next.js site, Supabase-authenticated practice app, and Express/Prisma API.

Next.jsSupabaseExpressPrisma

Built in production. Explained clearly.

My engineering work and 3,000+ hours of one-on-one teaching shape the same habit: make complicated systems understandable, then make them work.

At Flax, I closed silent data-loss paths across eight healthcare integrations, built a customer status page that removed support load during an outage, and shipped an AI-safety feature from Python evaluation through PostgreSQL to the TypeScript UI.

Flax.ai · Full-stack TypeScript Developer

Owned reliability and defect resolution across a HIPAA-regulated healthcare platform, shipping 131 pull requests across 12 packages.

DataAnnotation · AI Model Evaluator

Evaluate frontier coding models on real open-source codebases: write specifications, assess implementations, and document failure modes.

Focus Coding · Owner & Instructor

Run a tutoring business and its production software, from a Next.js marketing site to a Supabase-backed practice app.

BYU Family History Technology Lab · Research Assistant

Improved an unsupervised handwriting-recognition model through data preparation, augmentation, and PyTorch model iteration.

Tools I reach for.

A practical generalist stack, grounded in strict TypeScript, real tests, and an appetite for learning the systems around the code.

TypeScriptReactNext.jsNode.jsPythonPostgreSQLSupabasePrismaAWSDockerVitestAI evaluation

Get in touch

Let’s build something that holds up.

I’m open to software-engineering and AI-adjacent opportunities where careful technical work has an outsized impact.