Indie AI Lab — We explore emerging AI capabilities through rapid prototyping and real-world solutions

$ whoami
Independent AI researcher building tools that solve real problems

$ ls experiments/
production-systems/    developer-tools/    research-frameworks/    design-systems/

$ cat philosophy.txt
while (exploring) {
  prototype();
  test(); 
  iterate();
  ship();
}

The AT3 Stack Kit is a comprehensive monorepo with three specialized packages:

Key Features:

• AI-native stack built on Next.js 15, React 19, TypeScript, Tailwind CSS 4, and Supabase
• Edge-first architecture with serverless deployment patterns
• Intelligent project scaffolding and migration codemods
• Advanced development toolkit with integrated testing and linting
• Production-ready with comprehensive CI/CD workflows

Production Systems

• ML applications with real users and revenue
• Developer tools that ship and get adopted
• AI-enhanced productivity platforms
• Privacy-first local AI implementations

Research & Frameworks

• Inclusive design systems and methodologies
• Alternative productivity models for diverse teams
• Human-centered AI interaction patterns
• Open source tools for developer communities

Client Solutions

• Custom AI integrations for real-world problems
• Developer experience consulting and tools
• Accessibility and inclusive design implementation
• ML systems with economic data and complex APIs

AI/ML          │ Python, Scikit-learn, Local AI Models, OpenAI, Claude API
Backend        │ FastAPI, Node.js, Flask, SQLite, PostgreSQL, Azure, Supabase
Frontend       │ Swift (macOS), React, Next.js, TypeScript, CSS Variables, Tailwind CSS
Developer Tools│ CLI Design, Chrome DevTools Protocol, Git Integration, pnpm, Biome
Document Proc. │ AST Parsing, Static Analysis, JSON Schema, YAML, Document Processing
Data Sources   │ Eurostat APIs, ECB APIs, Economic Indicators, AST Analysis
Standardization│ YAML Schema Design, Platform Adapters, Universal Configuration
Infrastructure │ Azure Container Apps, Vercel, GitHub Actions, npm Registry, Edge Functions

Production Metrics

• 🎯 99.89% ML accuracy (FairRent rental price prediction)
• 📦 Multiple active npm packages (VDK CLI, AI Changelog Generator, AI Context Schema, Starlight Document Converter)
• 🌍 Live platforms serving real users
• 💰 €35-50/month operational costs for ML production systems
• ✅ 100% test coverage on production developer tools
• 🔄 10+ AI providers supported across toolchain

Research Impact

• 🏗️ Frameworks adopted by developer communities
• ♿ Accessibility-first design systems and methodologies
• 🧠 Neurodivergent-friendly productivity research
• 🔒 Privacy-first AI tool implementations
• 🎯 Universal AI compatibility across coding assistants
• ⚡ Edge-native development patterns and serverless architectures
• 📋 109+ curated AI development patterns across platforms

Open to:

• ML applications for real-world problems
• Developer productivity tool partnerships
• Inclusive design and accessibility consulting
• AI integration for privacy-conscious teams
• Modern stack adoption and migration consulting

Expertise areas:

• Economic data analysis and ML production systems
• macOS native development with AI features
• Chrome DevTools Protocol and browser automation
• Neurodivergent-friendly UX design and research
• AST analysis and code intelligence systems
• Universal AI assistant integration patterns
• TypeScript monorepo architecture and tooling
• Document processing and schema standardization

Not interested in:

• Generic chatbot implementations
• Blockchain/crypto pivots
• "AI-wash" marketing projects
• Breaking things for the sake of moving fast

Recent discoveries and build logs:

• Building VDK CLI: From prototype to production-ready developer tool with 311 tests and universal AI compatibility
• AI Changelog Revolution: From basic git logs to professional Keep a Changelog format with 10+ AI providers
• Building FairRent: From weekend hack to 99.89% accuracy production ML system
• AI Context Schema: Universal standardization for AI assistant behavior - solving configuration fragmentation across Claude Code, Cursor, Windsurf, and GitHub Copilot
• AT3 Stack Evolution: Modernizing T3 stack with AI-native patterns and edge deployment
• Shipping Developer Tools: npm package journey and MCP server implementation
• Document Processing Pipeline: Starlight Document Converter enabling seamless content migration
• Local AI Integration: Privacy-first approaches in macOS native apps
• Reverse Engineering Windsurf: Chrome DevTools Protocol adventures
• Inclusive Design Research: Energy-based workflows vs time management
• Economics APIs for ML: Lessons from Eurostat and ECB integration
• AST Analysis at Scale: Parsing 20+ languages for intelligent code analysis
• Monorepo Transformation: Building comprehensive toolkit ecosystems

• Lab Notes: [Coming soon - technical diary and research findings]
• Email: lab@entro314.com
• GitHub: @entro314labs
• Projects:
  • VDK CLI (npm) - Universal AI assistant toolkit
  • AI Changelog Generator (npm)
  • AI Context Schema (npm)
  • Starlight Document Converter (npm)
  • VibeCodingRules Platform (live)
  • AT3 Stack Kit (alpha monorepo)
  • Framework Code (GitHub)

┌─ entro314@labs ~/experiments
└─ $ git log --oneline --graph
* feat: ship VDK CLI v2.5.0 to production (100% test coverage)
* feat: ship AI Context Schema v2.1.0 - universal AI standardization
* feat: ship Starlight Document Converter to production
* feat: achieve universal AI assistant compatibility
* feat: ship AT3 stack kit alpha monorepo (3 packages)
* feat: ship FairRent ML to production (99.89% accuracy)
* feat: release AI changelog generator v2.5.1
* feat: launch VibeCodingRules platform  
* research: mountain path framework design
* feat: complete warm sunset design system
* wip: cascade backup tool for windsurf users
* wip: dotfiles manager with local AI

┌─ entro314@labs ~/experiments
└─ $ echo "Making AI less terrible, one weekend at a time"
Making AI less terrible, one weekend at a time

┌─ entro314@labs ~/experiments
└─ $ █

_{entro314 labs • Independent • Experimental • Pragmatic • Shipping Real Solutions}

Building AI tools that solve real problems, with a focus on developer experience, accessibility, and privacy-first approaches. From weekend prototypes to production systems serving real users.