███╗   ███╗███████╗██╗  ██╗███████╗██╗   ██╗███████╗██╗  ██╗
████╗ ████║██╔════╝██║  ██║██╔════╝██║   ██║╚══███╔╝██║  ██║
██╔████╔██║█████╗  ███████║█████╗  ██║   ██║  ███╔╝ ███████║
██║╚██╔╝██║██╔══╝  ██╔══██║██╔══╝  ██║   ██║ ███╔╝  ██╔══██║
██║ ╚═╝ ██║███████╗██║  ██║██║     ╚██████╔╝███████╗██║  ██║
╚═╝     ╚═╝╚══════╝╚═╝  ╚═╝╚═╝      ╚═════╝ ╚══════╝╚═╝  ╚═╝

Applied AI Engineer · Co-Founder @ Smartloop · Local AI Orchestration 🧠

20+ years shipping software. Microsoft MVP (C#). Currently deep in LLM fine-tuning, multi-agent systems, and local AI inference. Previously building infra @anaplan, dev-tools @telerik. Now building Smartloop — run AI agents on your own machine, connect tools, build custom skills, and get things done without sending your data anywhere.

• Building Smartloop — a local AI agent harness with an OpenAI-compatible API; run quantized open models on your hardware, connect MCP servers, and build custom skills, no cloud required
• Building agentic workflows to extract information and generate new content
• Fine-tuning local models and building LLM infrastructure
• Occasionally writing about LLM fine-tuning, evaluation, and agentic design patterns

• Post-Training: PEFT (LoRA/QLoRA), SFT, DPO, RLHF
• Evaluation: LLM-as-judge, offline eval harnesses, benchmark design
• Serving: OpenAI-compatible APIs, INT4/INT8 quantization, speculative decoding, distillation
• Agentic Systems: Multi-agent coordination, RAG, text-to-SQL, tool use
• Languages: Python, Go, TypeScript, C#

• 🤖 Smartloop — Local AI orchestration. Multiple small models, one smart router. Your data stays on your machine.

• 🌐: smartloop.ai
• 𝕏: @mehfuzh
• 💼: linkedin.com/in/mehfuzh
• 📧: mehfuz@gmail.com

"Think → Build → Repeat."

_{I have a cat named KitKat 🐱 · Puedes hablarme en español}