AI Projects
Explore my portfolio of artificial intelligence projects. You can find the source code for most of these projects on my GitHub.
View on GitHubaicache
An AI CLI session caching system designed to enhance developer workflow efficiency by caching responses from various AI CLIs (Gemini, Claude, Qwen, OpenAI, gcloud, llm).
Target Audience:
Developers using AI CLIs
Why We Built This:
To improve developer productivity by caching AI CLI responses.
Key Features:
- Advanced caching
- Intelligent argument parsing
- Persistent storage
- Comprehensive cache management
- Interactive tools
- Shell integration
- Semantic search
- Behavioral analytics
- Predictive prefetching
- Multi-modal support
- Enhanced security
Technology:
Python
AI Education Platform (aiteach)
An innovative AI/ML education platform offering personalized learning paths based on user proficiency levels. It integrates educational content from leading academic institutions and industry leaders.
Target Audience:
Students and professionals looking to learn AI/ML
Why We Built This:
To provide a personalized and effective learning experience for AI/ML education.
Key Features:
- Proficiency assessment
- Adaptive course recommendations
- Content with progressive difficulty
- Level-based organization
- Secure JWT-based authentication
- Full CRUD for courses and lessons
- Progress tracking
- Gamification
- Community forum
- Responsive design
Technology:
Backend: FastAPI (Python) with PostgreSQL and Docker. Frontend: Flutter
Classic Games - World-Class Cross-Platform Gaming Suite (classic-games-monorepo)
A unified monorepo featuring graphics-rich implementations of classic card and board games with realistic 3D graphics and online multiplayer.
Target Audience:
Casual gamers and enthusiasts of classic board games
Why We Built This:
To create a high-quality, modern, and accessible platform for playing classic games.
Key Features:
- Poker (Texas Hold'em)
- Backgammon
- Scrabble
- Realistic 3D graphics with physics
- Cross-platform (iOS, Android, Web)
- Online multiplayer with matchmaking
- Real-time synchronization
- Offline AI opponents
- Sound effects
- Haptic feedback (mobile)
- Responsive design
Technology:
Frontend: React, React Native, Next.js, TypeScript, Three.js, React Three Fiber, Expo GL. Backend: Node.js server with WebSocket support. Monorepo management with Turborepo.
Echo - Audio Memory Platform (echo)
A mobile-first application that intelligently and passively records audio, using AI to transcribe, identify speakers, and tag key moments automatically. It aims to make memories searchable and relivable for generations.
Target Audience:
Individuals who want to capture and cherish their memories
Why We Built This:
To create a new way to capture, organize, and relive audio memories.
Key Features:
- One-tap ambient recording
- AI transcription with speaker diarization
- Chronological Echo Stream
- Universal search
- Memory curation
- Secure sharing
- Privacy-first
Technology:
Backend: Node.js with Express, PostgreSQL with full-text search, Google Cloud Platform (Cloud Storage, Speech-to-Text API, Cloud Natural Language API), JWT authentication. Frontend: React (planned).
Little Canvas - World's #1 Kids Super App (littlecanvas)
A comprehensive kids' application for ages 3-8+ that combines coloring, storytelling, learning modules, rewards, and offline activities in a safe, engaging platform. It's enhanced with AI-powered personalization, global localization, and advanced safety features.
Target Audience:
Children aged 3-8+ and their parents
Why We Built This:
To provide a safe, engaging, and educational digital playground for children.
Key Features:
- Advanced creative tools
- AI-powered learning system
- Global localization
- Ultimate safety & privacy
- Offline-first architecture
- Comprehensive learning system
- Advanced gamification & rewards
- Creative storytelling platform
- Seasonal content
- Offline activity integration
- Inclusive accessibility
- Social & family features
- Enhanced parent dashboard
Technology:
Frontend: React Native for cross-platform (iOS & Android) with Redux Toolkit. Backend: Firebase for authentication and data storage, serverless functions, AI/ML components for personalization.
NexusComm - The Ultimate Unified Communication Platform (nexuscomm)
A comprehensive unified communication platform that consolidates all personal, professional, and social digital communications into one intelligent, organized stream, aiming to eliminate app-switching fatigue.
Target Audience:
Busy professionals and individuals who want to streamline their communication
Why We Built This:
To solve the problem of app-switching fatigue and create a more unified communication experience.
Key Features:
- AI-powered intelligence
- Unified communication
- Enterprise-grade security
- Business intelligence
- Accessibility
Technology:
Microservices architecture. Backend: Node.js 18+ with Express.js, PostgreSQL with TypeORM, Redis, Socket.IO, Bull.js, OpenAI GPT models. Frontend: React Native with Expo (mobile), React with TypeScript (web). DevOps: Docker, Kubernetes-ready, GitHub Actions.
P2P Gear Rental Platform (p2p-gear-rental)
A modern peer-to-peer marketplace for renting photography and videography equipment, enabling users to list their gear and discover equipment from other creators in their area.
Target Audience:
Photographers, videographers, and content creators
Why We Built This:
To create a community-driven marketplace for renting creative equipment.
Key Features:
- Gear listings
- Advanced search filters
- Rental management
- User profiles
- Secure payments via Stripe
- Supabase-powered authentication
Technology:
Frontend: Next.js 15, React 19, TypeScript, Tailwind CSS. Backend & Database: Next.js API Routes, Prisma ORM, PostgreSQL, Zod. Authentication & Payments: Supabase, Stripe. Development & DevOps: Jest, Playwright, ESLint, Docker, GitHub Actions.
Universal Agent Builder (UAB)
A web-based application for creating AI agents from prompts, supporting A2A (Agent-to-Agent Protocol), ADK (Agent Development Kit), and MCP (Model Context Protocol) frameworks.
Target Audience:
Developers and researchers working with AI agents
Why We Built This:
To simplify the process of creating and configuring AI agents.
Key Features:
- Interactive agent creation
- Support for multiple AI frameworks
- Pre-built templates
- Auto-framework selection
- Real-time configuration forms
- Built-in mock testing environment
- State persistence
- Export functionality
- Multiple deployment configuration options
Technology:
Vanilla JavaScript for the web application.
Vibe Coding Charter (vibecharter)
A comprehensive governance framework designed to transform AI from a spontaneous automaton into a disciplined, architecture-aware co-pilot, addressing the challenge of harnessing AI's speed without sacrificing long-term architectural integrity.
Target Audience:
Software development teams using AI assistants
Why We Built This:
To provide a framework for using AI assistants in a disciplined and architecturally sound way.
Key Features:
- Four core architectural principles
- Five non-negotiable rules
- Layered defense mechanism
- Prevention of abstraction gaps
- Advanced prompt engineering guidance
- Introduction of an 'Intent Engineer' role
Technology:
Primarily a conceptual framework and documentation project, not a software application with a distinct tech stack for its implementation.
Lucent (NextGenMediaManager)
A cutting-edge iOS application that revolutionizes media organization through comprehensive AI-powered features, including duplicate detection, semantic search, photo clustering, and intelligent insights. All processing is on-device for complete privacy.
Target Audience:
iPhone users with large photo libraries
Why We Built This:
To bring powerful, on-device AI to photo management on iOS, ensuring privacy and performance.
Key Features:
- Intelligent tagging
- Smart collections
- Natural language search
- OCR text recognition
- Semantic embeddings
- NIMA-powered aesthetic quality scoring
- Duplicate and near-duplicate photo detection
- K-means photo clustering
- ML insights
- Seamless CloudKit integration
- Modern SwiftUI interface
Technology:
iOS (SwiftUI, Photos Framework, CloudKit, Combine, Core Location, Vision Framework). External Services & ML Models: Google Gemini AI, Google MLKit, CLIP Embeddings, NIMA Model, custom ML models. Development: macOS 14+, Xcode 15+, iOS 17+ SDK.
WellnessApp - Enterprise-Grade Wellness Platform (wellness)
A comprehensive, production-ready wellness application with a secure Node.js backend API and a native iOS client. It empowers users to manage their mental, physical, and social wellbeing holistically.
Target Audience:
Why We Built This:
Key Features:
- Backend: JWT authentication, user management, APIs for mood/sleep/workout/nutrition tracking, meditation library, habit system, real-time messaging (WebSocket), e-commerce (Stripe), file uploads (Google Cloud Storage), security (rate limiting, input sanitization), and email service.
- iOS App: Guided meditation, mood/stress tracker with AI insights, breathing exercises, journaling, sleep/activity tracking, HealthKit integration, custom habit tracker, AI wellness assistant (personalized nudges, insights), community features (forums, challenges), biometric authentication, and personalized onboarding.
Technology:
Backend: Node.js 18+ with Express.js, MongoDB 5.0+, Socket.IO, Stripe, Google Cloud Storage, Nodemailer. iOS: iOS 16.0+, Swift 5.9, SwiftUI, MVVM architecture, Alamofire, Socket.IO Client, Core Data (encrypted), Keychain Services, HealthKit.