Skip to content
PORTFOLIO.EXE :: FULLSTACK.EXPLORER

SYSTEM READY | ROUTES INDEXED | STATIC EXPORT ONLINE

$ boot

Raq Robinson

Applied AI Software Engineer

I build AI powered products from idea to shipped app.

My recent work spans agentic systems, pricing workflows, audio ML pipelines, real time collaboration, and accessible frontend platforms. I care about making complex technology feel usable, reliable, and clear.

I use AI tools heavily, but I do not treat generated code as finished code. I validate the work with tests, evals, failure mode coverage, and live product review.

$ status

  • Shipped AI products and full stack prototypes
  • Agent workflows, ML pipelines, and real time systems
  • Evaluation, testing, and failure mode coverage
  • Accessible frontend systems and design systems
  • Production deployment with Docker, Railway, FastAPI, Node, Redis, and WebSockets

$ query how_i_work

How I work

In practice, that means I own the quality of the output, not just the prompt.

  1. 01Turn the product goal into requirements and failure cases
  2. 02Build the smallest testable version
  3. 03Add evals, unit tests, and manual review gates
  4. 04Instrument latency, cost, errors, and edge cases where relevant
  5. 05Ship, inspect the live app, and document what changed

$ query featured_system

Two connected music AI products

Oh Sheet turns songs into engraved piano sheet music. TuneChat turns those scores into multiplayer practice rooms.

Together, they show how I build full stack AI products: pipeline orchestration, async jobs, fallback handling, real time state, browser audio, product UX, and shared design systems.

Oh Sheet home screen: paste a YouTube link to get piano sheet music

Oh Sheet

Paste a YouTube link and get engraved piano sheet music.

Oh Sheet runs a five stage pipeline that pulls audio, transcribes it, arranges it for piano, and exports PDF, MusicXML, and MIDI files. Built with FastAPI, Celery, Redis, Docker, Railway, and ML transcription tools.

What it shows: AI pipeline design, async processing, external API integration, fallback handling, and production deployment.

[Case study][Live site] (opens in a new tab)[GitHub] (opens in a new tab)
  • Python 3.12
  • FastAPI
  • Celery
  • Redis
  • yt-dlp
  • Docker
  • Railway
  • AGPL-3.0
TuneChat room with sheet music, chat, and a playable piano keyboard

TuneChat

TuneChat is a multiplayer piano room in the browser.

Players can join a shared room, view the same score, practice with a virtual piano, and stay synced through Socket.io. Built with Node, Express, Socket.io, Tone.js, OpenSheetMusicDisplay, and vanilla JavaScript.

What it shows: real time collaboration, shared state, browser audio interaction, music notation rendering, and product focused UI.

[Case study][Live site] (opens in a new tab)
  • Node.js
  • Express
  • Socket.io
  • Vanilla JS
  • Vite
  • Tone.js
  • OpenSheetMusicDisplay
  • tcalgo
  • MuseScore CLI
[All projects][About me][Download resume]