Welcome

How do you want to read this?

Choose how much depth you want. You can switch anytime.

Software Engineer & Creative Designer

Hi,
I build for you.

I build software and systems. From the user-facing app to the infrastructure underneath.

What I've built.

Just Smash That Just Smash That

01

Just Smash That

Software & Creative Lead

A rage room startup that needed everything. Reservations, staff ops, a brand identity, and an online presence. I joined a 4-person team pre-launch and built the software and creative infrastructure from the ground up. Handled all digital marketing and advertising, taking us from 15K monthly website visitors to over 300K. Created all the content, ran the campaigns, and managed the brand across every platform. We started as a single rage room and have since expanded to three additional activities in the same facility, grew from 4 to 40+ employees, and are now opening a second location.

The problem

The founders had a vision but no infrastructure. They were booking customers through a third-party, managing staff on paper schedules, and trying to build a brand with no design system or marketing strategy.

Business impact

The business grew from a single rage room to four activities under one roof, 40+ employees, and a second location on the way. Website traffic went from 15K to 300K monthly. Every employee works from the same system, from the front desk to the backend labor cleaners.

The problem

Xola owned availability data but not the cart, pricing, or bundle rules. Third-party APIs can't be called from the client, and business logic tied to an external service is brittle — it can't evolve on its own schedule. SEO had no technical foundation: no structured data, no programmatic targeting, no schema strategy.

How it was solved

Xola became read-only. Availability flowed in; our schema controlled everything after — cart state, pricing, bundles, upsells. Clean separation meant the checkout experience could evolve without touching the booking engine. The SEO system drove 15K to 300K monthly visitors. Paid campaigns used actual booking conversion data, not vanity metrics.

Applied principles

Build for the floorSimplicity over feature densityRestraint in designShip the whole thing
View details
Next.jsSupabaseFigmaIllustratorGoogle Ads
Planning
Processing
Processing
Processing
Polish
Architectural Visualization Architectural Visualization

02

Architectural Visualization

Blender & Unreal Engine

Produced contractor-grade interior renders for the Daytona Beach location. Took a raw floor plan and turned it into a full 3D space with modeling, lighting, and materials. These renders didn't just look good, they were functional. They were used to secure interior wall layouts and act as the visual blueprint for the entire build crew, preventing costly mistakes during the framing and electrical phases by communicating the vision before a single wall was built.

The problem

The contractor needed a way to communicate interior design intent to investors, builders, and the crew. They wanted to avoid expensive staged furniture setups and vague sketches.

Business impact

The renders became the visual contract for the build. Framing and electrical crews had a reference that prevented miscommunications and rework. Investors got a pitch-quality presentation that helped secure funding.

The problem

Concept renders aren't useful on a construction site. The contractor needed spatial accuracy — wall placements, electrical routing, door clearances — not just something that looked good. Dimensionally incorrect renders would cause miscommunications and rework during framing and electrical phases that cost more than the renders were worth.

How it was solved

Every wall, doorway, and electrical path matched contractor-verified measurements. Lumen's real-time GI made light behavior accurate enough to drive interior material and placement decisions before construction started. Investors got a pitch-quality presentation. Builders had a shared reference that prevented the most expensive kind of mistake: one nobody caught until the walls were already up.

Applied principles

SimplicityPerformance-firstProgressive enhancement
View details
BlenderUnreal Engine 5AutoCADPhotoshop
Planning
Planning
Planning
Planning
Processing
Processing
Processing
Processing
Processing
Processing
Processing
Polish
Polish
Polish
Clyra Clyra

03

Clyra

HR & Operations Platform

Built to replace the paper schedules, group texts, and spreadsheet payroll that small businesses are still running on. Clyra gives owners, managers, and employees one system for scheduling, timesheets, payroll, and inventory without the bloat of enterprise HR software. What started as a side project for one shop became a mission-critical platform that has handled thousands of shifts and millions in payroll across multiple locations.

The problem

Small businesses are still running on paper schedules, group texts, and spreadsheet payroll. It's fragmented, error-prone, and everyone is constantly playing catch-up. From the owner to the newest employee, nobody has a single source of truth.

Business impact

Owners can run scheduling, timesheets, and payroll from one place. Employees get clarity on their shifts and can swap without texting the whole team. It's handled thousands of shifts and millions in payroll across multiple locations.

The problem

Standard auth assumes one person per device. Clyra runs on shared kiosks where 20 employees use the same screen. Multi-tenant isolation had to sit below the application layer — a buggy API route can't be allowed to leak one location's data into another. 264 endpoints also meant UI state management had to be systematic or inconsistency would compound across the entire surface.

How it was solved

PIN-based tuple auth resolves identity in a single event and times out automatically for kiosk mode. RLS at the Postgres layer means isolation is enforced even when application code is wrong. A shared hook pattern makes loading, error, and cache state consistent across 264 endpoints. Three-tier caching keeps the platform fast without managing cache invalidation by hand.

Applied principles

Clean architectureDRYBuild for the floorYAGNIRestraint
View details
Next.jsReactSupabaseTailwind
Planning
Planning
Processing
Polish
Polish
Polish
Vesper Vesper

04

Vesper

Education Technology

Early Development

Architecture education platform designed specifically for CTE high school classrooms. Vesper solves the problem of teachers only seeing a final 3D model without knowing the steps taken to build it. It introduces process-aware tracking, essentially giving teachers a playback reel for student builds. This allows for real accountability and ensures that the student actually understands drafting principles rather than just copy-pasting assets or following rote tutorials.

The problem

Teachers only see the final 3D model. They have no idea if the student actually learned the drafting process or just copied someone else's work. It's impossible to assess understanding when you can't see the process.

Business impact

Teachers get a playback reel of every build step. Every primitive, every boolean operation, every decision. They can actually grade for understanding, not just output. Students have to show their work.

The problem

Three packages that had to evolve independently without coupling. Scene state, 3D rendering, and interactive tools all needed clean boundaries. Timeline recording had to be meaningful — only real architectural decisions, not UI noise. And the editor had to be usable by a CTE high school student with no prior CAD experience.

How it was solved

Each package owns its layer and nothing else. @pascal-app/core can be used without a viewer. The viewer can render without editor tools. Tools can be swapped without touching the scene graph. The Go backend's append-only timeline lets teachers replay every student decision from the first action. The AI reads build state and suggests a next step but never executes one — the learning is in the decision, not the output.

Applied principles

Build for the floorSimplicityClean architecturePerformance-first
View details
ReactTypeScriptSupabaseZustand
Planning
Planning
Planning
Planning
Processing
Polish
Polish
Polish
Polish
Digital Boards Digital Boards

05

Digital Boards

Display Systems

Customer lobby boards designed to show packages, entry cost, and rules in a digital menu format. High visibility displays that update automatically so the front desk never has to worry about outdated signage.

The problem

The business needed a way to display their packages, pricing, and house rules to customers in the lobby. They were using printed signs that got outdated constantly and needed manual replacing every time prices changed.

Business impact

Every customer walking in sees the current packages and prices instantly. No more outdated signs. The digital menu shows packages, entry costs, and rules all in one place. The front desk doesn't have to answer "how much is it?" a hundred times a day.

The problem

A static sign is infrastructure nobody maintains. Every price change, package update, or rule revision required someone to physically replace the sign. The solution needed to propagate changes instantly, survive reboots without touching it, and require zero ongoing maintenance.

How it was solved

Any admin update reaches every lobby screen in real time over WebSocket. The Pi costs $35 and runs indefinitely without intervention. The display has no user-facing interaction — just a render loop. Nothing to break, no state to manage, no maintenance cycle.

Applied principles

SimplicityRestraintBuild for the floor
View details
Next.jsWebSocketsRaspberry Pi
Planning
Processing
Polish
Axe Branding Axe Branding

06

Axe Branding

Brand Identity

Complete brand identity for a client, from concept to final execution. Built a visual language that matched the energy and edge of the product.

The problem

A mark that only works with gradients and glow doesn't work in print, embossed, or monochrome. The glyph had to carry energy and edge on its own, at low fidelity, in black on white — before any treatment was applied.

How it was solved

Procreate sketches forced the concept to stand alone. Photoshop pressure-tested it across real production contexts. If it failed in any condition, the concept was wrong. Final Illustrator vectorization locked in the angles and line weights — the mark works without effects.

View details
ProcreatePhotoshopIllustrator
Planning
Processing
Polish
Hoodie Branding Hoodie Branding

07

Hoodie Branding

Merch Design

Merchandise design for a brand. From sketch to final product.

The problem

DTG printing loses edge fidelity on fabric. Fine lines and tight curves that look clean on screen break down when the print driver interprets them. The design had to be engineered toward the output constraint from the start — not retrofitted for it after the work was done.

How it was solved

Digital garment mockups in Photoshop validated proportion and contrast at wear scale before any refinement happened. Final Illustrator vectorization produced clean edge geometry that survives the print driver without degradation.

View details
ProcreatePhotoshopIllustrator
Planning
Processing
Polish
Logo Design Logo Design

08

Logo Design

Brand Identity

Logo and brand identity design. From initial concepts to production-ready assets.

The problem

A logo has to survive everywhere: business card, billboard, dark mode, favicon, embroidery. Most fail because they were designed for one context and break everywhere else. The glyph had to hold the brand at any scale, in black and white, before type or color were ever introduced.

How it was solved

The mark was validated small and monochrome before any treatment was added. Photoshop tests across real-world contexts confirmed it held up. Final Illustrator system delivers a specification — clear space rules, misuse examples, type lockups — not just a logo file.

View details
ProcreatePhotoshopIllustrator
Planning
Processing
Processing
Polish
Paint Branding Paint Branding

09

Paint Branding

Brand Identity

Brand identity for a client in the paint industry. Built a visual system that matched the precision and craft of the trade.

The problem

The mark had to survive physical production formats: labels, van wraps, uniforms. Color accuracy wasn't optional. A mark that looked right on screen but failed at physical output was useless — and the formats varied enough that each one needed explicit validation.

How it was solved

Geometric constraint in the sketch phase produced forms that felt precise without being clinical. Physical format mockups ran before refinement to validate contrast and scale under real conditions. Final Illustrator assets met color accuracy requirements for physical production output.

View details
ProcreatePhotoshopIllustrator
Planning
Processing
Polish
VR Branding VR Branding

010

VR Branding

Brand Identity

Brand identity for a VR company. Modern, futuristic aesthetic that felt like it belonged in the experience.

The problem

The instinct for VR is always gradients, glow, and holographic effects — all of which date fast and fail without color treatment. The mark had to feel dimensional and future-facing on its own. If it only worked with effects, it wasn't a strong enough mark.

How it was solved

Geometric forms validated in dark environments with holographic lighting found what held up without effects. Final vectorization built for multi-platform delivery — the mark works in AR environments, on hardware, and on screen without needing treatment to survive any of them.

View details
ProcreatePhotoshopIllustrator
Planning
Processing
Polish

Your business runs on duct-taped workflows. I replace them with real systems.

Origin

When I was twelve I started jailbreaking iPhones. Not because it was cool, but because the phone didn't do what I wanted it to do. So I made it. I dug into system files, broke things, fixed them, and learned how software actually works by forcing it to work for me.

That's still what I do. The difference is now it's for businesses. You've got schedules on paper, payroll in a spreadsheet, and half the team communicating through group texts. I look at all of that and build one system that handles it. Something clean, something everyone on the team can actually use without calling somebody for help.

Replace, don't patch

If the process is broken, adding a feature on top of it doesn't fix anything. I find the thing that's actually failing and rebuild it. Fewer steps, less confusion, actually works.

Build for the floor, not the ceiling

The person who just got hired last week should be able to open the app and use it. If they can't, it's not done. I always build for the least technical person in the room.

Ship the whole thing

I don't just write code and hand it off. I do the engineering, the design, the branding, the infrastructure. One person who sees the full picture moves way faster than a team playing telephone.

Korvix.
The products.

Korvix is my operating name. It's the umbrella under which I build and own software products.

Clyra

HR & Ops

Multi-tenant HR and operations platform. Timesheets, scheduling, payroll, inventory, RBAC, kiosk clock-in, and audit logging. Built for small teams that need real infrastructure without enterprise overhead.

264 API endpoints

57 pages

$5.25M FMV

Vesper

Education

Architecture education platform for CTE classrooms. Process-aware modeling where teachers can replay student build history, not just the final model. Think AutoCAD stripped down to what actually matters for learning.

In active development

Korvix CLI

Developer Tools

Internal CLI for scaffolding projects, managing deployments, and automating devops tasks. Saves hours every week across the studio.

Internal use

Focused on agentic workflows and what comes next.

Right now the focus is infrastructure. I'm running self-hosted AI agents that handle development and project management on their own. Eva handles engineering, Jarvis handles operations. They run on a Linux machine accessed via Tailscale, with n8n orchestrating task flows and Discord as the control plane.

Clyra is moving toward an LLM integration layer that replaces the dashboard with a conversational interface. Vesper is in active product development.

React Node.js Supabase Linux LangGraph n8n Tailscale Blender

Let's work
together.

Got a project in mind? I'd love to hear about it.