AI-native engineering · Claude Code · Real users

Justin Scott

AI-Native Full-Stack Engineer

I build and ship real software by directing AI coding agents — and bring the system design, UX judgment, and verification that turn generated code into products people actually use.

Software engineer at Idemia-NSS (Angular, Spring Boot, PostgreSQL) by day. The rest of the time I ship my own products with Claude Code — and build the skills and MCP tooling that make it faster.

Tampa, FL · Remote-friendly

downloads of my Minecraft mods
7,500+
users on my election-results site this year
5,000+
npm downloads of my open-source MCP server
200+/mo
shipping enterprise software at Idemia-NSS
4+ yrs
Featured work

Things people actually use

Live products with real users — each one designed, built, and shipped end-to-end.

US Election Tracker — 2024 electoral map with the 270-to-win bar and state-by-state resultsLive

US Election Tracker

5,000+ users this year

Live election-night results with a poll-closing schedule, prediction-market odds for every major race, and House and Senate control odds. Plus state-by-state electoral maps for all 25 presidential elections since 1928, an 11-layer demographics map, 51 state hubs, and two quizzes.

Next.jsTypeScriptInteractive SVG mapsPrediction-market oddsCensus & BLS dataNetlify
Fabric · Modrinth
Minecraft Fabric Mods
Live

Minecraft Fabric Mods

7,500+ downloads

Quality-of-life Fabric mods, kept current with each Minecraft release through a Claude Code pipeline I built: a custom skill, an MCP decompiler to diff versions, and my own publishing MCP server.

JavaFabricMixinsGradleClaude CodeMCP
Avery Light FX — Halloween light show site with show hours and the FM station for musicLive

Avery Light FX

Visitor info site for a holiday light show

Info site for a drive-up Halloween and Christmas light show: hours, show status, directions, and Venmo donations for charity. Includes a secured admin panel wired to the show’s xSchedule API.

Next.jsTypeScriptNextAuth.jsxSchedule APITailwind CSSNetlify
How I work

AI does the typing. I do the engineering.

Anyone can prompt a model. The leverage is knowing what to ask for, catching what it gets wrong, and getting the result in front of real users. That's the job I do with Claude Code every day.

  1. 01

    Frame the problem

    Specs with hard boundaries and a definition of done before any code: what is in scope, what must not change, and how we will know it works.

  2. 02

    Design the system

    The calls an agent shouldn't make alone: data model, failure modes, rate limits, caching, and what happens under real traffic.

  3. 03

    Direct the agents

    Claude Code with custom skills, MCP servers, and parallel subagents — and when the tooling I need doesn't exist, I build and publish it.

  4. 04

    Verify like a skeptic

    Real browsers at phone and desktop widths, both themes, production-like conditions. Small, revertable commits. Done means proven.

In practice · US Election Tracker, Oct 2026

Three weeks before election night I took the site from audit to production in one session: 80 baseline screenshots across phone, desktop, light, and dark; a server pinned to UTC that exposed a hydration bug hitting every US visitor; a stale election-night header and an independent candidate drawn as a Republican, both caught before launch. It shipped as 21 small, individually revertable commits with before/after evidence for every change.

Experience

Where I've worked

Enterprise delivery in regulated industries, plus the infrastructure work that keeps teams fast.

  1. Software Engineer · Idemia-NSS

    Jun 2024 – Present

    Tampa, FL · Remote

    • Ship full-stack features in Angular, Spring Boot, and PostgreSQL for enterprise clients in highly regulated industries, where reliability, security, and auditability are non-negotiable
    • Cut technical debt ~40% in three months by modernizing legacy Java services and tightening code standards
    • Improved application performance ~30% through Spring Boot microservice design, PostgreSQL query tuning, and a reactive Angular frontend
    • Use AI tools (Cursor, Claude) to move faster on spikes, refactors, and exploration, with every production change still going through team review, testing, and release
  2. Software Engineering Intern · Idemia-NSS

    Mar 2022 – Apr 2024

    Morgantown, WV · Onsite

    • Rebuilt the Jenkins CI/CD pipeline as modular, reusable components, cutting build, test, and deploy times 70% and making pipelines easier to maintain and onboard onto
    • Modernized multi-platform mobile apps to current Android and iOS SDKs
  3. Software Engineering Intern · River Point Technology

    May 2021 – Aug 2021

    Washington, PA · Hybrid

    • Wrote Terraform infrastructure-as-code for AWS, halving the time to stand up a new environment
    • Automated continuous delivery to staging and production with GitLab CI/CD pipelines
    • Built an MkDocs engineering knowledge base of code patterns and runbooks for new contributors
Skills

What I build with

AI-native workflow on top of a real full-stack foundation — the second is what makes the first trustworthy.

AI-native engineering

Claude Code (skills, subagents, hooks)MCP server developmentPrompt & context engineeringAgentic workflows (LangGraph)RAGAnthropic & OpenAI APIsCursor

Product & system design

System designReal-time systems (WebSockets)API designUX & accessibility (WCAG)Performance & Core Web VitalsTechnical SEO

Languages

TypeScriptJavaPythonJavaScriptSQLBash

Frameworks

Next.jsReactSpring BootAngularNode.jsFastAPIReact Native / ExpoTailwind CSSFabric (Minecraft)

Data

PostgreSQLSQLitepgvectorRedisPinecone

Cloud & DevOps

AWSGCPNetlifyDockerTerraformGitHub ActionsJenkinsGitLab CILinux
Education

Education

Bachelor of Science in Computer Science

May 2024

West Virginia University · Morgantown, WV

Dean's ListScholarship of DistinctionAWS Senior CapstonePenetration Testing
Contact

Let's talk

Always up for talking shop or the right opportunity. Based in Tampa, FL — open to remote and mostly-remote hybrid.

jscottwv@gmail.com · GitHub