Gary Hyde

Open to senior and lead roles
Senior Full-Stack Developer
gary@jobs.garyhyde.com
View my workGet in touch

Things I have shipped

A few things I have built and shipped. Two of them are live, so go ahead and click through.

Pathara landing page with the headline "Your job search, handled"
AI ProductLive
Pathara
A complete job search service powered by an AI agent pipeline

Pathara turns a resume and a short questionnaire into a ranked list of real open jobs, each with a tailored resume, cover letter, and interview prep doc. I built the intake flow, the AI agent pipeline that does the research and writing, and the per-candidate results portal with downloads and an application tracker.

->AI agent pipeline from intake to finished documents
->Per-candidate portal with downloads and an application tracker
->Supabase auth, row-level security, and storage end to end
Next.jsTypeScriptSupabaseAI agentsStripeVercel
Kora direct booking engine landing page
PaymentsLive
Kora
Direct booking engine for service businesses

A direct booking engine that lets service businesses take reservations on their own site instead of handing a marketplace a 20 percent cut. I built the booking and scheduling flow, Stripe payment processing with Stripe Tax scoped from day one, guest communications, and a review pipeline.

->Booking, scheduling, and payments in one engine
->Stripe and Stripe Tax scoped from day one
->Guest communications and a review pipeline
Next.jsNode.jsExpressPostgreSQLStripeStripe Tax
Outlandr.io landing page showing the island settlement game
GameLive
Outlandr.io
Free browser multiplayer island-settlement strategy game

A real-time multiplayer settlement game that runs entirely in the browser, with no download and no account needed. I built the Phaser rendering, the real-time multiplayer for two to six players, seeded board generation, fog of war, turn timers, and bot opponents that can fill empty seats.

->Real-time multiplayer for two to six players
->Phaser engine rendering hex tiles at 60fps
->Bot opponents, fog of war, and seeded boards
Next.jsTypeScriptPhaserWebSocketsVercel
Lumi home page showing a natal chart and AI reading
AI ProductLive
Lumi
AI astrology app that turns a birth chart into a personal reading

Lumi calculates a precise natal chart from your birth details using Swiss Ephemeris astronomy, then has an AI write a reading that actually sounds like you instead of a template. I built the chart engine, the freemium product flow, and the AI reading pipeline, and shipped it on Vercel.

->Swiss Ephemeris chart math for observatory-grade accuracy
->AI readings generated per person, not from a template
->Freemium flow: a free Big Three, a paid full reading
Next.jsTypeScriptOpenAI APISwiss EphemerisVercel
TeeTimes course search and booking page
Proof of ConceptLive
TeeTimes
Demo: multi-tenant tee time booking engine built as a collaboration prototype

A proof-of-concept booking platform for golf country clubs, built as a demo for a potential collaboration. Not a product being offered. I designed the multi-tenant data model, the booking engine, and the background job pipeline to validate the architecture. The core booking and scheduling patterns from this PoC later shaped Kora.

->Engine-first multi-tenant architecture as a PoC
->Background jobs on BullMQ and Redis
->Booking patterns that later informed Kora
Next.js 14ExpressPostgreSQLDrizzle ORMAuth.js v5BullMQRedisTailwind CSS
Illustration of three warehouses syncing stock through a central node
E-commerceShipped
Multi-Location Inventory for WooCommerce and Odoo
Stock and order sync across locations between WooCommerce and Odoo

A custom integration that keeps stock and orders in sync across multiple warehouse locations between a WooCommerce storefront and an Odoo ERP. I built the sync pipeline and the custom Odoo module work, and tracked down cross-system inventory bugs at the data layer, including a multi-location overselling issue that spanned both systems.

->Multi-location inventory sync across two systems
->Order pipeline between WooCommerce and Odoo
->Root-caused a cross-system overselling bug
WordPressWooCommercePHPOdoo 17PythonREST APIs
Luma Loop live event photo capture and projection display
Custom PluginShipped
Luma Loop
Real-time photo capture and projection for a live venue

A custom WordPress plugin for a live venue client. Staff capture photos on a tablet, the images poll in over AJAX, and a projector display updates in real time. I scoped, documented, and delivered the whole thing end to end for a paying client.

->Tablet capture to projector in real time
->AJAX polling refresh loop
->Scoped and shipped solo for a paying client
WordPressPHPJavaScriptAJAX

The stack

Frontend
React
Next.js 14
TypeScript
Tailwind CSS
shadcn/ui
SASS / SCSS
Backend
Node.js
Express
PostgreSQL
Drizzle ORM
Redis
BullMQ
REST APIs
AI & Tooling
AI-augmented development
AI agent workflows
AI code review
Multi-model synthesis
Prompt engineering
Commerce & CMS
WordPress
WooCommerce
Shopify
Stripe
Stripe Tax
Odoo 17

Models I have tested

I keep a rotation of models and pick the right one for the job. Here is where each one earns its place in my workflow.

Claude
Anthropic
Coding + agentic

The model I trust most for real code and long agentic work. It holds a large codebase in context, sticks to a plan, and is the engine behind most of my day-to-day building in Claude Code.

ChatGPT
OpenAI
General reasoning

My quick second opinion. I reach for it to rubber-duck a design, draft copy, or reason through a tradeoff before I commit to an approach.

Grok
xAI
Fast + current

Good when I want a fast answer with a current view of the web. I use it for speed and recency rather than deep deliberation.

Kimi
Moonshot
Long context

My pick when a task needs a huge context window and a strong agentic researcher. It works through large document sets and multi-step research without losing the thread.

Nemotron
NVIDIA
Open weight

An open-weight model I have tested for reasoning work. Useful when I want a capable model I can run on my own terms instead of behind a closed API.

Ollama
Local runtime
Local + private

How I run models locally when a task should never leave my machine. Great for private codebases, offline work, and cheap iteration with no API bill.

Composer
Cursor
In-editor agent

Cursor's in-editor agent. I use it for fast, in-context edits while I am already in the flow, where staying inside the editor beats switching tools.

On AI-assisted development

Short notes on how I actually work with these tools. I treat AI as a force multiplier, not a replacement for judgment, and most of the leverage lives in the workflow around the model.

NOTE 01
gstack and agent-first workflows

Garry Tan's gstack reframes the coding agent as a workflow, not a chat box. Skills for planning, review, QA, and ship turn one-off prompts into a repeatable pipeline. The leverage is in the process around the model, not just the model.

WorkflowAgents
NOTE 02
Superpowers: skills as reusable expertise

Obra's Superpowers packages real engineering discipline (test-driven development, systematic debugging, planning) into skills the agent invokes on its own. It is the difference between an agent that guesses and one that follows a method.

SkillsDiscipline
NOTE 03
Harness engineering

Most of the quality in AI-assisted work comes from the harness: the tools, verification steps, and context discipline wrapped around the model. I spend as much time building that harness as I do prompting. The model is the engine, the harness is the car.

ToolingQuality
NOTE 04
Loops and routines

Loops and scheduled routines let an agent keep making progress between my check-ins: watching a deploy, sweeping for bugs, iterating until it hits a target. Used with care, they turn a one-shot tool into something that works while I focus elsewhere.

AutomationAgents

Who I am

I am a senior full-stack developer based in the Orlando metro, Florida, with 10+ years shipping production web applications. I have built multi-tenant SaaS platforms, payment and booking infrastructure, e-commerce systems, and custom tooling for clients. I own work end to end: from schema design and API through to a polished, fast frontend.

Over the last two years I have built with AI in the loop as a core part of how I work. I use Cursor and Claude Code daily. I also built my own multi-model code review tooling that synthesizes output from several models into a single Ship/Fix/No-Ship scorecard, which has tightened feedback cycles and caught real bugs before they reach production.

I work best on small, senior teams where I can own a meaningful surface. I care about production quality, approachable codebases, and fast feedback loops. The code I ship is the code I would want to read six months later.

Gary Hyde
14+
Years in production
12+
Years building independently
6+
Projects shipped end to end
AI-native
AI-augmented workflow

Get in touch

I am open to senior and lead full-stack roles, remote-first or around the Orlando metro. Tell me what you are building and what you need, and I will reply within one business day.

Prefer email? gary.robert.hyde@gmail.com