🔮 Founder @ Wizardry Labs — AI Product Studio (100/100 Google Lighthouse Web Performance) ⚡ Low-Level AI Infra → C++17 & Apple Metal Paged KV-Cache Engine (vLLM block paging + NVMe SSD offloading, 90% RAM reduction) 🛍️ Multimodal AI → Storefront AI (Shopify Vision & OCR Assistant), Pakistan Food Intelligence Platform, IslandIQ (macOS App) 🤝 Open to full-time roles & remote engineering contracts — burhanmoindev@gmail.com

Zero-dependency inference engine for Apple Silicon M-series chips. Implements virtual memory block page tables (16 tokens/block) with dynamic NVMe SSD offloading, enabling 100K+ token context processing on 16GB Macs without memory thrashing or OOM crashes. Includes custom Metal Shading Language compute kernels (kernel_gemma4_rmsnorm, GEMV) and an ANSI terminal TUI telemetry visualizer.

National menu & restaurant intelligence platform indexing 500,000+ data entries for sub-millisecond geospatial retrieval using SQLite FTS5 full-text search and grounded local Ollama LLMs. Studio site (wizardrylabs.burhanmoin.com) achieves perfect 100/100 Google Lighthouse scores across Performance, SEO, Accessibility, and Best Practices.

Full-stack conversational e-commerce shopping assistant for Shopify storefronts. Features multimodal Vision & OCR image processing for instant customer photo product matching, async worker queues, and automated fallback LLM routing.

High-concurrency gaming platform backend handling 100,000+ concurrent users. Replaced $O(n^2)$ bottlenecks with an $O(n)$ set-based draw calculation algorithm for sub-second result computation, dual-wallet atomic ledger, and custom GC memory daemon.

Native macOS dynamic notch overlay assistant bringing screenshot vision, persistent AI chat, media controls, and system actions into an always-available desktop interface.

📊 Stats Overview