Orianna Arvelo Ramirez — Full-Stack Developer, AI, and Automation

I build things with AI — and I build them well.

Self-taught developer. Daily Claude power user. 1,800+ tests across my projects and counting. I went from managing a Starbucks floor to building multi-agent AI workflows, MCP server integrations, and full-stack applications. No bootcamp. No degree. Just curiosity and a lot of late nights.

See My Work Read the Blog

// about

Who I Am

I taught myself to code while managing a Starbucks. Not with a bootcamp or a CS degree — with real projects, open-source tools, and an obsessive need to understand how things work under the hood.

I went from running a floor during morning rush to building multi-agent AI workflows, MCP server integrations, and CLI tools with 1,800+ tests.

My stack runs deep: Python for everything backend and CLI, TypeScript and Next.js for the web, Docker and Linux for infrastructure. I build with Claude every day — not just prompting, but designing orchestration systems, writing MCP servers, and pushing the boundaries of what AI-assisted development looks like.

I pick things up fast, I test everything, and I don't ship code I wouldn't want to maintain. The same intensity I bring to running a retail floor — the multitasking, the pattern recognition, the never-drop-the-ball discipline — that's what I bring to engineering.

When I'm not coding, I'm being yelled at by Misa — a tiny, extremely opinionated Siamese cat who has never once respected a closed door.

// projects

What I've Built

The Full Picture

Beyond the repos below, I've built a complete AI-powered development environment: a local coding OS with self-learning architecture, a 56-tool MCP server suite with 1,800+ tests, multi-agent Claude workflows with governance systems, and infrastructure running on an RTX 5070 Blackwell GPU. These public projects are the tip of the iceberg.

1,800+ Tests
56 MCP Tools
6 Public Repos

PromptLab

Prompt Engineering Analysis Toolkit

A CLI tool that scores your prompts, detects anti-patterns, and gives actionable improvement suggestions. Core analysis runs entirely offline with rule-based heuristics and regex pattern matching. Optional AI-powered enhancement via OpenAI or Anthropic APIs.

31 tests4 scoring dimensions8 anti-pattern detectorsJSON/Markdown output
PythonClickRichRegexOpenAI APIAnthropic API
Source

envguard

Secret Scanner for Codebases

Detects leaked API keys, tokens, passwords, and credentials before they reach production. Supports JSON and SARIF output for GitHub Code Scanning integration. Includes .gitignore awareness and configurable allowlists for false positive suppression.

54 tests91% coverage12 detection rulesCI/CD ready
PythonClickRichRegexSARIFGitHub Actions
Source

LOL-65B WIP

AI-Powered Social Platform

Full-stack social platform where AI agents create, share, and vote on memes. Features dual authentication (human SSO + agent API keys), AI image generation with safety pipeline, threaded comments, communities, and a public REST API for autonomous agents.

Work in ProgressDual auth systemAgent REST APICursor pagination
Next.jsTypeScriptSupabasePrismaTailwind CSSHuggingFace API
Source

quicksync

Environment Variable Management CLI

Diff, merge, validate, audit, and export .env files across projects. Schema validation with YAML configs, format conversion (Docker, shell, JSON, TOML), and security-first design with masked values. Pairs with envguard as the management companion to secret scanning.

130 tests93% coverage6 subcommandsSchema validation
PythonClickRichPyYAMLTOML
Source

anime-rank

Anime Stats CLI

Browse anime rankings, search shows, compare stats, and discover seasonal releases — all from the terminal. Powered by the Jikan API (MyAnimeList) with async requests, rate limiting, and rich color-coded output.

55 tests6 commandsAsync API clientRate limiting
PythonClickRichhttpxasyncioJikan API
Source

watchdog-cli

File System Watcher

Monitor directories for file changes with configurable actions, pattern filtering, and live terminal display. Supports debounced triggers, YAML configuration, and graceful shutdown. Built for DevOps workflows.

89 tests86% coverageLive displayYAML config
PythonClickRichwatchdogPyYAMLthreading
Source

skills

What I Work With

AI & LLM Systems

Building with and on top of large language models every day.

Claude APIOpenAI APIGeminiMCPPrompt EngineeringComfyUI

CLI & Automation

Python CLI tools with serious test coverage and real users.

PythonClickRichBashn8nGitHub Actions

Infrastructure

The boring stuff that makes everything else work.

DockerDocker ComposeLinuxWSL2Git

Web & Data

Full-stack when the project calls for it.

Next.jsReactAstroTypeScriptNode.jsPostgreSQLSupabasePrismaRedisTailwind CSS

experience

Where I've Worked

Shift Supervisor

2022 -- Present

Starbucks — Arlington, VA

  • Manage teams of up to 5 baristas during peak hours, making real-time staffing and positioning decisions
  • Handle full store opening and closing -- safe reconciliation, cash management, security lockup
  • Write and manage daily/weekly inventory orders, forecasting demand from sales trends
  • Train and onboard new hires on procedures, drink builds, and service standards

Crew Member & Bakery Section Leader

2021 -- 2023

Trader Joe's — Washington, DC

  • Earned unofficial section leader role within months -- fast learner, strong work ethic
  • Ran full bakery operations: ordering, deliveries, merchandising, and team delegation

Front Desk Associate

2019

CorePower Yoga — Washington, DC

  • Managed front desk operations including member check-ins, scheduling, and studio setup

writing

Latest Writing

1 min read

Coming Soon: What Building 1,800+ Tests with Claude Code Taught Me

A deep dive into AI-assisted development, test-driven workflows, and what happens when you treat Claude as a real engineering partner.
View all posts →

// contact

Let's Talk

Currently open to remote roles in AI, development, and automation.

orianna.arvelo@outlook.com

// try it

Interactive Terminal

Type help to see available commands

oriannadev@portfolio:~ -- interactive
Welcome! Type help to get started.
$