Featured Projects
A showcase of full-stack applications demonstrating modern engineering practices, security implementations, and technical leadership.
Enterprise-grade TypeScript library for persistent AI agent memory management using PostgreSQL with pgvector. Solves the critical problem of conversation continuity across sessions with intelligent compression, multi-model support, and zero-cost local embeddings.
Technology Stack
Key Features
- •Zero-cost embeddings with local Sentence Transformers
- •Multi-model token counting for OpenAI, Anthropic, DeepSeek, Google, Meta
- •Intelligent memory compression with 4 compression strategies
- •Sub-5ms memory operations with pgvector similarity search
- •Universal tokenizer with provider-specific optimizations
- •Production-ready with comprehensive Docker test suite
Architecture Highlights
- •PostgreSQL-native design with pgvector for vector operations
- •ULID-based ID generation achieving 28M operations/second
- •Hybrid compression system with automatic summarization
- •Type-safe validation with Zod schemas
- •ESM/CJS dual distribution with full TypeScript support
redlock-universal
Production-Ready Distributed Redis Locks
Open Source Infrastructure Library • Published on NPM
Enterprise-grade distributed Redis lock library implementing the Redlock algorithm with universal client support. The first and only library supporting both node-redis and ioredis natively, designed for production microservices and distributed systems.
Technology Stack
Key Features
- •True Redlock algorithm with clock drift handling
- •Universal support for node-redis v4+ and ioredis v5+
- •Zero runtime dependencies for maximum reliability
- •<1ms lock acquisition in 99th percentile
- •Comprehensive error handling and retry mechanisms
- •283 tests with 95%+ coverage (unit/integration/e2e)
Architecture Highlights
- •Adapter pattern for universal client compatibility
- •Consensus-based distributed lock acquisition
- •TypeScript-first with strict mode compilation
- •Performance-optimized with memory leak prevention
- •ESM/CJS dual package for maximum compatibility
deploy-mcp
Universal Deployment Tracker for AI Assistants
Developer Tools & AI Integration • Published on NPM
Track deployments across all platforms directly in your AI conversation. Built for the Model Context Protocol (MCP), it brings deployment status directly into AI assistants like Claude Desktop, VS Code, Cursor, and more. No more context-switching or dashboard hunting.
Technology Stack
Key Features
- •Universal AI support via Model Context Protocol
- •Real-time deployment badges with webhook integration
- •Secure local-only API token storage
- •Lightning-fast <2s deployment status checks
- •Multi-platform support (Vercel, Netlify coming)
- •Zero data collection - privacy-first approach
Architecture Highlights
- •MCP server for AI assistant integration
- •Cloudflare Workers for global edge performance
- •TypeScript-first with strict type safety
- •Modular adapter pattern for platform extensibility
- •ESM/CJS dual package distribution