Challenge

Career-matching tools tend to demand lengthy questionnaires or return generic, unexplained suggestions. The goal was a fast, free path from ‘here’s who I am’ to ranked roles with real reasoning, with zero signup friction for a reviewer trying it cold.

Build

Built in Next.js App Router with real Gemini 2.5 Flash integration, not a mock: a four-question chat wizard or resume upload feeds a structured profile, roles are ranked with reasons via generateObject, and a live-streamed AI pitch is drafted for the top 3 matches. Supabase adds an optional upgrade path — magic-link or Google sign-in graduates anonymous localStorage data into a Postgres backend with row-level security.

Outcome

A complete AI product loop — profile extraction, ranked matching with reasoning, and a streamed personalized pitch — that works end-to-end for anonymous visitors with no paywall, deployed live on Vercel.

Challenge

ISP signup flows are usually a maze of forms split across pages, with no sense of progress and a jarring handoff between marketing site and checkout. The goal was a single continuous flow from ‘is fiber available here’ to a confirmed install date.

Build

Built in React, TypeScript, and Vite with Tailwind CSS v4: a marketing site (hero, plans, testimonials, FAQ) feeding into a five-step modal wizard — address verification, plan and add-on selection, install scheduling, account creation, and confirmation — driven by shared component state with Framer Motion transitions between steps.

Outcome

A cohesive, single-session path from landing page to a confirmed installation slot, with live form validation and a polished order-confirmation summary.

Challenge

Most portfolio landing pages default to generic SaaS templates. The goal was a concept that could carry its own visual identity — bold editorial type, a warm wine-and-gold palette, and tango-inspired motion — while staying accessible and fast.

Build

Built in Next.js 16 App Router with React 19, TypeScript, and Tailwind CSS 4: an abstract CSS hero with ambient, reduced-motion-aware animation, scroll-triggered Intersection Observer reveals, and an interactive event discovery section with category filters, live text search, and save/unsave states, backed by local demo data since this is a frontend-only concept.

Outcome

A polished, responsive marketing site and event-discovery UI that demonstrates editorial design sense and interaction detail without a real backend behind it.

Challenge

Kitchens are typically run on paper tickets and memory. The goal was to give a kitchen the same clarity a manager gets from a SaaS dashboard: live, legible, and uncluttered, without adding another screen to operate.

Build

Built in Next.js App Router and TypeScript: a live ticket board (New → Fire → Plating → Ready), a ticket detail panel, station load and chef ownership views, and a service intelligence panel, all driven by shared React state and a lightweight motion layer.

Outcome

A six-module command center — Command, Intake, Prep, Inventory, Insights, and Crew — built as a single connected product rather than a series of separate screens.

Problem

Slow KPI loading, delayed revenue charts, laggy search, slow filters, and sluggish dashboard refresh.

Optimization

Removed artificial delays, memoized revenue calculations, optimized search/filtering, and reduced table rendering.

Result

Render time improved from 2000ms to 140ms with smoother dashboard interactions.

Problem

Laggy search, slow filters, heavy rendering, expensive ticket processing, and poor responsiveness.

Optimization

Memoized dataset generation, optimized filtering, added debounced search, deferred updates, and pagination.

Result

Render time improved from 3675ms to 2.3ms with smoother search, filtering, and table interactions.