Flagship project

Valorcy

AI Website Builder SaaS Platform

Valorcy is a complete SaaS ecosystem for building, launching, and managing websites with AI. It is not a single site — it is one product delivered as three production applications that share a design system, type definitions, utilities, and authentication inside a single monorepo.

Production applications: 3Production applications
Shared monorepo: 1Shared monorepo
Feature modules: 40+Feature modules
Languages supported: 3Languages supported

Visit live site

One product, three applications

Each application is built and deployed separately, but they share the same foundation — so they look, feel, and behave like one product.

Marketing WebsiteThe public front door

User DashboardThe application customers live in

Admin PanelRunning the platform

Shared foundationShared monorepo

One component library consumed by all three applications — a button, dialog, or data table is defined once and behaves identically everywhere
A single design system of tokens, typography, and spacing, so the marketing site, dashboard, and admin panel are unmistakably the same product
Shared TypeScript types spanning the database schema, API contracts, and UI props — a schema change surfaces as a compile error, not a runtime bug
Shared utilities for formatting, validation, permissions, and data access, removing duplicated logic across apps
Unified authentication and session handling — users move from the marketing site into the dashboard without a second login
Common architecture and coding standards that keep three applications reviewable and maintainable by a small team

Marketing Website

The public front door

Visit live site

https://valorcy.com

Purpose

The public-facing surface of the platform: landing pages, pricing, template gallery, the AI website builder showcase, blog, documentation, and SEO-focused content. It is the first thing a prospective customer sees and the entry point into authentication, so it is tuned for speed, discoverability, and conversion.

Key features

Content & pages

Landing pages
Pricing
Template gallery
AI Website Builder showcase
Blog
Documentation

Growth & performance

SEO-focused pages
Dynamic metadata
Core Web Vitals tuning
Server-side rendering
Authentication entry
Responsive design

Challenges solved

Keeping rich animation and heavy visual content from degrading Core Web Vitals — solved by server rendering the critical path and deferring motion until after first paint.
Making template and blog content indexable while it is still driven from the same database the dashboard writes to, using incremental static regeneration instead of a separate CMS export.
Explaining an AI product visually without a video-heavy page that costs seconds of load time.

My contributions

Built the marketing application end to end, from page architecture to the final motion pass.
Implemented the dynamic metadata layer so every template, pricing tier, and blog post ships correct titles, descriptions, and Open Graph tags.
Tuned Core Web Vitals — image strategy, font loading, and render-blocking work — to keep the site fast on mobile.
Wired the authentication entry point that hands users off into the dashboard with an established session.

Technologies

Next.jsTypeScriptTailwind CSSFramer MotionSSR / ISR

User Dashboard

The application customers live in

User Dashboard

Purpose

The primary SaaS application and the one customers interact with daily. It is where they generate a site with AI, edit it visually, manage content and commerce, connect a domain, and run their subscription. Every other surface of the product exists to support what happens here.

Key features

Build

AI Website Builder
Drag & Drop Editor
Themes
Templates
Design System

Content

CMS
Blog Management
File Manager
Multi-language support

Commerce

Products
Orders
Analytics

Platform

Website Management
Domains
Integrations

Account

Team Members
Subscription Management
Billing
Settings
Authentication

Challenges solved

A drag-and-drop editor has to stay responsive while holding an entire site tree in memory — solved with granular state slices and memoized subtrees so a single block edit does not re-render the canvas.
AI generation is slow and non-deterministic; the interface had to stream progress and stay usable rather than blocking behind a spinner.
One dashboard has to serve a user with a one-page portfolio and a user running a storefront with orders and a team, without either seeing an interface built for the other.
Multi-language content management meant modeling translations at the data layer, not bolting a language switcher onto the UI.

My contributions

Built the core builder and editor experience, including the block model, the drag-and-drop canvas, and the property panels behind it.
Implemented the AI website generation flow end to end — prompt interface, streaming feedback, and conversion of the AI response into editable blocks.
Developed the CMS, blog, products, and orders modules against a shared relational schema.
Built subscription, billing, and team-member management, including the plan gating that controls which features a workspace can reach.
Owned the reusable component layer this dashboard is assembled from — the same layer the admin panel consumes.

Technologies

Next.jsTypeScriptTailwind CSSShadcn UISupabaseDrizzle ORMTanStack Query

Admin Panel

Running the platform

Admin Panel

Purpose

The internal application used to operate the SaaS: managing users and their subscriptions, defining plans, moderating published websites, curating templates and content, configuring the AI, and controlling who inside the company can do what.

Key features

People

User Management
Roles & Permissions
Support Tools

Revenue

Subscription Management
Plan Management

Content

Website Moderation
CMS Management
Blog Management
Templates Management

Platform

Analytics Dashboard
AI Configuration
Feature Flags
Audit Logs
Settings

Challenges solved

Internal tools touch every record in the system, so access had to be genuinely role-based — enforced at the data layer rather than by hiding buttons in the UI.
Support staff need to see a customer's workspace to help them, without that becoming an unlogged path into private data — resolved with scoped impersonation and audit logging.
Plan and feature-flag changes take effect across the dashboard immediately, which meant designing them as configuration the other app reads, not as code deployments.

My contributions

Built the admin application on the shared component library, so it inherited the product's design language rather than looking like a bolted-on internal tool.
Implemented roles and permissions along with the audit logging that records every privileged action.
Developed plan and subscription management, including the feature-flag system the dashboard reads to gate functionality.
Built the moderation and template curation workflows the team uses to keep published content and the gallery healthy.

Technologies

Next.jsTypeScriptTailwind CSSShadcn UIPostgreSQLDrizzle ORMRBAC

What this demonstrates

Designing and shipping a multi-application product rather than isolated websites
Building a scalable component system that holds up across three very different UIs
Owning features end to end — data model, API layer, interface, and polish
Balancing marketing-grade performance and SEO with dashboard-grade interactivity
Working with role-based access, billing, and moderation — the parts of SaaS that only appear at production scale

Next.jsTypeScriptTailwind CSSShadcn UISupabasePostgreSQLDrizzle ORMFramer MotionTurborepo