Khalid Alnujaidi

Computer Scientist & Systems Architect

[RSS Feed]

Focus on high-level system architecture, computer vision infrastructure, embedded hardware engineering, and autonomous AI orchestration layers.

Core Technical Highlights

// Systems Engineering & Machine Learning

  • Designed and deployed a production-grade MultiModal LLM fraud detection system using vision models to parse refund images, scaling directly into regional delivery infrastructure — adopted by Delivery Hero →.
  • Built LeastGen — a transparent inference optimizer that routes 95.1% of agent queries to local cache at 0ms latency, saving 715M tokens on a single GPU, using two VRAM-pinned specialized models (3B Planner + 4B Executor) — LeastGen →.
  • Bootstrapped an AI B2B venture serving 10+ clients across public sector, private sector, and banks — 300k SAR revenue at a 66% average profit margin. Among its projects: a drive-through loyalty system using automatic license-plate recognition, developed to prototype.

// Hardware & Rapid Prototyping Stack

  • Programmed custom firmware and hardware integrations across ESP32 and Raspberry Pi environments for IoT automation and modular robotics.
  • Structured multi-part mechanical enclosures and rapid-prototyping iterations optimized for structural shells using Bambu Lab ecosystems.
  • Built a DIY mini-supercomputer — one reusable USB image that turns recycled PCs and laptops into a self-healing k3s cluster (52 vCPU / 82 GB), with offline autoinstall, Tailscale mesh, and one-command joins — DIY Mini-Supercomputer →.