Local AI / LLM Mentor — Help Me Set Up and Understand AI Running Locally

I'm looking for an experienced local-AI/LLM developer who can spend a few hours with me over 1 to 2 live screen share sessions teaching me how to run AI locally on my Windows laptop. I'm specifically interested in understanding the technology, not simply having someone install everything for me. I'd like you to explain what we're doing as we go and leave me with enough understanding that I can continue experimenting on my own afterward. My hardware includes an ASUS ROG Zephyrus with: - CPU: AMD Ryzen AI 9 HX 370 - RAM: 32 GB LPDDR5 - GPU: NVIDIA GeForce RTX 4070 Laptop GPU (8 GB VRAM) - Integrated GPU: AMD Radeon 890M - NPU: AMD Ryzen AI NPU - Storage: ~2 TB NVMe SSD - OS: Windows I'd like to learn how to make practical use of this hardware for local AI, particularly the NVIDIA GPU, while also understanding what the CPU, RAM, and NPU can and can't contribute. What I'd like to accomplish: - Evaluate my hardware and determine what size/type of models I can realistically run - Set up a local LLM environment (likely Ollama, but I'm open to your recommendation) - Teach me how local inference actually works at a high level - Show me how to download, manage, configure, and interact with different local models - Explain things like model size, quantization, context windows, VRAM/RAM requirements, GPU acceleration, etc. - Set up a user-friendly interface if appropriate (e.g. Open WebUI or something similar) - Explain how local models differ from cloud models - Ideally set up an autonomous/agentic AI system that can run locally on my laptop - Teach me how agents work, including tools, permissions, memory, and how an agent can interact with my computer - Help me understand the security/privacy implications of giving a local agent access to my files, terminal, browser, etc. - Give me some hands-on exercises/examples so I can learn how to experiment with this myself I'm expecting this to be a short consulting/teaching engagement rather than a large software-development project — probably around 2–4 hours to start. I'm looking for someone who has real hands-on experience with: - Ollama or similar local inference platforms - Local/open-weight LLMs - NVIDIA CUDA and GPU-accelerated inference - AI agents / agentic workflows - Tools such as LangChain, LangGraph, Open WebUI, MCP, CrewAI, AutoGen, or similar - Running models on consumer hardware - Windows, WSL2 and/or Docker Python - Securely sandboxing agents and controlling their access to the local system Teaching ability is important to me. Please don't just tell me what you can install — tell me about your experience teaching people how these systems work. In your proposal, please include: - What local-AI stack you'd recommend for my specific hardware and why - What models you'd recommend starting with and why - What agent framework(s) you have hands-on experience with - An example of something you've helped someone run locally - How you would structure a 2–4 hour teaching session - Your hourly rate I'd prefer a live, interactive session where I can ask questions and understand what we're doing rather than having the work done asynchronously. Important: I want to learn how to build and manage this system myself. Please don't simply configure everything for me without explaining what you're doing.