Abdul Moiz Shahzad

Toronto, Canada|moizibnyousaf@gmail.com|github.com/MoizIbnYousaf|linkedin.com/in/abdulmoiz-shahzad|moizibnyousaf.com

Summary

Self-taught full-stack AI engineer with 5 years of experience building agent infrastructure and shipping production apps. Built an open source AI agent skill installer (1,006 stars, 12K+ npm installs). Developed RAG-powered geospatial search for a mobile app serving 500+ users.

Experience

Ai-Agent-Skills [npx ai-agent-skills]

Dec 2025 – Present

Creator & Maintainer · Open Source

  • AI coding agents had no universal skill installer. Built and shipped the first one the day agent skills became an open standard, a month before skills.sh existed. Reached 1,006 stars and 12K+ npm installs in 4 months.
  • Wrote a full terminal UI for browsing, installing, and curating 64 skills across 47 repos and 12+ agent runtimes (Claude Code, Codex, Cursor). One command replaces manual per-runtime configuration.
  • Evolved the CLI into a curation layer: shelves, provenance tracking, and tooling for teams to build and share their own agent skill libraries. 107 forks, 15+ contributors submitting PRs and skill packages.

Halaali

2024 – Present

Co-Founder & Sole Engineer · Mobile App (iOS + Android)

  • One engineer, full product. Solo-built the full-stack React Native app, Supabase edge functions, PostgreSQL schemas, Stripe payments, admin portal, and marketing site. Shipped to both app stores, grew to 500+ users.
  • Built Habibi, an in-app AI agent for halal discovery using Vercel AI SDK and Claude. Tool-calling architecture with geospatial search, deals, and flyer lookup. RAG-powered semantic search across 1,200+ locations using pgvector and OpenAI embeddings. Adapts recommendations to user location, prayer times, and cuisine preferences.
  • Built a geospatial data layer indexing 1,200+ restaurants across 2 countries with PostGIS, raw PLpgSQL, and Mapbox. Real-time location queries with WebSocket live updates and push notifications.

SkillCreator.ai

Dec 2025 – Present

Founder · Desktop + Web

  • No good tooling existed for writing and testing AI agent skills. Built a local-first desktop IDE with a 6-dimension evaluation framework (correctness, tool selection, instruction adherence, latency, token efficiency, error recovery) across 8 agent runtimes.
  • Electron app with Monaco editor, SQLite for offline-first performance, Redis caching, and OpenTelemetry tracing for agent execution observability. Next.js web companion with GraphQL API.

Notable

Technical Skills

Languages: TypeScript, JavaScript, Swift, Kotlin, Python, SQL, PLpgSQL, HTML/CSS
Frameworks & Libraries: React Native, Expo, Next.js, Electron, Node.js, Express, Hono, Tailwind CSS, LangChain
Infrastructure & DevOps: Docker, AWS (EC2, S3), Nginx, GitHub Actions, Vercel, EAS, Redis, WebSockets, GraphQL
Data & Auth: PostgreSQL, PostGIS, pgvector, Supabase, SQLite, Stripe, JWT, OAuth 2.0
AI & Agent Tooling: LLM, Claude Code, Claude Agent SDK, Vercel AI SDK, MCP Protocol, Anthropic API, OpenAI API, RAG, Vector Search, Codex
Spoken Languages: English (fluent) · Urdu (native) · Arabic (conversational)