AI Development Projects Madison WI | Portfolio

SkillEngine

AI Career Productivity Platform — A full-scale SaaS application built in 31 days using AI-assisted development with Claude Code.

Live Application

Days to Build

125K+

Lines of Code

277+

AI Skills

App Pages

Workflows

Role Templates

311

Git Commits

Monthly Hosting

The Problem

Most people use AI the same way every time: open ChatGPT, explain what they need, get a response, close the tab. Tomorrow they start over with zero memory of what worked. Valuable prompts are lost, and professionals waste time re-explaining context repeatedly.

The Solution

SkillEngine saves your best prompts as reusable Skills — complete with expert personas, structured outputs, and industry-specific context. Run them with one click instead of re-explaining yourself constantly. Build once, use forever.

User Experience Flow

Home Dashboard

Welcome screen with value proposition, "How it works" flow, and key benefits.

Skill Library

Browse 236+ production-ready skills with role filters and featured collections.

Skill Runner

Setup checklist, AI provider selection, and dynamic input forms for each skill.

SkillEngine execution results showing Company Research Report output

Real-Time Results

Structured output with execution metrics, token usage, cost tracking, and export options.

SkillEngine Job Application Package workflow with 5 automated steps

Workflow Runner

Multi-step workflows combining skills like Job Application Package with 5 automated steps.

SkillEngine Settings showing API key management for multiple AI providers

Settings & API

Multi-provider API key management for Google Gemini, Anthropic Claude, and OpenAI.

Core Capabilities

🎯 For Job Seekers

Job Readiness Scorer — Match analysis for roles

Resume Customizer — Tailored bullet points

ATS Keyword Optimizer — Pass screening

Interview Prep — STAR-format answers

Salary Coach — Negotiation guidance

💼 For Professionals (27 Roles)

Software Engineer — Code review, specs

Financial Analyst — Reports, forecasts

Marketing Manager — Campaigns, strategy

Project Manager — Status, risk assessment

+ 22 more roles with 4-12 skills each

🏢 Enterprise Operations

Contract Review — Risk identification

Compliance — Program development

AI Governance — Implementation

Tech Debt — Assessment & planning

RFP Response — Automation

Technical Architecture

Multi-Provider AI Interface

┌─────────────────────────────────────────┐ │ Unified AI Interface │ ├─────────────┬─────────────┬─────────────┤ │ Gemini │ Claude │ ChatGPT │ │ • 2.0 Flash│ • Haiku │ • GPT-4o │ │ │ • Sonnet │ • 4o-mini │ │ │ • Opus │ • o1 │ └─────────────┴─────────────┴─────────────┘ Switch providers with one click. Model-specific token limits and streaming handled automatically.

Local-First Data Architecture

┌─────────────────────────────────────────┐ │ User's Browser │ │ ┌─────────────────────────────────────┐│ │ │ IndexedDB (Primary Storage) ││ │ │ • Workspaces • Skill history ││ │ │ • Custom skills • Preferences ││ │ └─────────────────────────────────────┘│ └─────────────────────────────────────────┘ │ Optional sync ▼ ┌─────────────────────────────────────────┐ │ Supabase (Cloud Sync) │ │ • Community skills • User profiles │ └─────────────────────────────────────────┘

Technology Stack

Frontend

React 18, TypeScript

Styling

Tailwind CSS

Build

Vite

Routing

React Router

Local DB

IndexedDB

Backend

Supabase

Hosting

Netlify

Testing

Vitest

AI-Assisted Development with Claude Code

Daily Development Velocity

📝

4,000+ lines of production code per day

🎯

9 new AI skills created daily

📦

10 git commits per day average

📄

1.5 application pages completed daily

Commit Distribution by Week

Week 1

Core architecture

Week 2

114

Skills + Roles

Week 3

Workflows

Week 4

Polish + Keys

"The cost and time equation for building software has fundamentally changed."

This project — 48 pages, 277 skills, 24 workflows, full authentication, multi-provider AI integration — would traditionally require a team of 3-4 developers working 2-3 months. Built solo in 31 days. Hosting cost: $0/month. The tools are here. The question is who learns to use them effectively.

Project completed December 2025

Resume Intelligence Search

A fully functional RAG (Retrieval-Augmented Generation) application built from scratch in under 2 hours using Claude Code — demonstrating just how fast AI-assisted development can turn ideas into working software.

<2 hrs

Total Build Time

RAG

Architecture

~500

Lines of Code

LanceDB

Vector Database

Claude

AI Synthesis

FastAPI

Backend

Build Timeline

Requirements + Scaffold 15 min FastAPI project + LanceDB setup

Embedding pipeline 20 min Sentence-transformer → LanceDB table

Search endpoint + Claude synthesis 25 min /search API with RAG retrieval

Minimal HTML/JS UI 30 min Full working prototype

Testing + bug fixes 20 min Production-ready state

Total 110 min Fully functional RAG app

Application in Action

Home Interface

Clean, simple search interface for natural language queries against resume data.

Search Output

RAG-synthesized answer with context retrieved from the resume vector database.

Resume search with percentage match scores

Match Analysis

Detailed matching with percentage relevance scores for retrieved resume sections.

Technology Stack

Backend: FastAPI (Python)
Vector DB: LanceDB (embedded, ~10MB)
Embeddings: sentence-transformers (all-MiniLM-L6-v2)
Synthesis: Claude API (claude-3-haiku)
Frontend: Vanilla HTML/JS

Traditional Build Estimate

Data modeling + API design → 4-8 hours
Embedding pipeline + testing → 4-6 hours
Search logic + integration → 4-6 hours
UI + polish → 4-8 hours
Total: 16-28 hours (2-3 days)

Aspect	Traditional Approach	AI-Assisted (Claude Code)
Time to MVP	2-3 days	<2 hours
Library selection	Manual research, trial & error	Claude recommends optimal stack
Boilerplate	Copy-paste from docs	Generated + tested instantly
Debugging	Stack Overflow + trial & error	AI pinpoints & fixes
Verdict	Slow	10-15x faster

Production Path

Prototype (Current) COMPLETED

Local LanceDB
Single-file codebase
Basic HTML/JS UI
Claude API key in env

Hosting: $0 (runs locally)

SaaS Version POTENTIAL

Hosted vector DB (Pinecone/Weaviate)
User auth + multi-tenant
React/Tailwind UI
Stripe integration

Estimated build: 2-3 additional days with Claude Code

"The barrier to building AI-powered applications has collapsed."

What previously required specialized ML expertise and days of development can now be built in under 2 hours. This isn't about replacing developers — it's about removing friction between ideas and working software.

Project completed January 2026

Projects & Case Studies

SkillEngine

Resume Intelligence Search

SkillEngine

The Problem

The Solution

User Experience Flow

Core Capabilities

🎯 For Job Seekers

💼 For Professionals (27 Roles)

🏢 Enterprise Operations

Technical Architecture

Multi-Provider AI Interface

Local-First Data Architecture

Technology Stack

AI-Assisted Development with Claude Code

Daily Development Velocity

Commit Distribution by Week

Resume Intelligence Search

Build Timeline

Application in Action

Technology Stack

Traditional Build Estimate

Production Path

Prototype (Current) COMPLETED

SaaS Version POTENTIAL

Explore More Projects

Have a Project in Mind?