I am not a developer by trade. I had never written a single line of Python before this project.

What I had was a problem to solve, a curiosity about AI, and enough stubbornness to figure it out one error message at a time.

I started with zero knowledge of Flask, APIs, WebSockets, databases, or machine learning. I learned everything you see in my repositories through trial, error, research, and rebuilding things that broke. Every concept — routing, blueprints, async threads, vector databases, EMA crossovers, IMAP protocols — was brand new to me when I started.

Personal Neuro Layer is a fully working personal AI assistant I run in my own home. Not a demo. Not a tutorial project. A real system I use every single day.

It manages my:

• 🏠 Smart home via Home Assistant
• 📧 Email, documents, contacts
• 🔐 Network security with auto-banning
• 🎵 Music through Music Assistant
• 🌐 MikroTik router and QNAP NAS
• 💰 Cardano crypto trading with live signals and paper trading
• 🧠 Long-term memory with vector search
• 🎤 Wake word listener and voice interface
• 📱 Mobile face UI accessible via Cloudflare tunnel
• 🔍 Web search and live news
• ✍️ Self-writing code review system

The architecture — a central routing layer with specialized modules hanging off it — mirrors what the AI industry is now recognizing as the right way to build personal AI systems. I arrived at it independently by just asking "how would a brain work?"

Partly curiosity. Partly because I wanted something that truly worked for my life.

And partly in memory of my daughter Kelly-Ann Fleming, who passed away in 2019 at 27 years old. One of my earlier projects — KAF Coin — was created in her honor with the goal of helping families who lose loved ones cover unexpected costs. That purpose still drives everything I build.

If you found this repo you are looking at proof that you don't need a computer science degree, a coding bootcamp, or years of experience to build something real.

You need a problem worth solving and the will to keep going when it breaks.

Personal Neuro Layer is open source. Take it, run it, make it yours.

LM Studio is required to run the AI brain locally. Performance depends entirely on your hardware.

Recommended minimum specs:

• RAM: 16GB (32GB recommended)
• GPU: NVIDIA GPU with 8GB+ VRAM for best performance
• CPU: Modern 6-core or better (CPU-only mode works but is slow)
• Storage: 20GB+ free for models

The default model used is Qwen2.5-14B — a strong balance of intelligence and speed on mid-range hardware. If your machine struggles you can swap to a smaller 7B model in the config with no other changes required.

No cloud API required — everything runs locally on your machine.

• 📱 React Native mobile app
• 🏢 Small business client portal
• 🌐 Hosted service for anyone who wants their own Jarvis

Built by a first-time Python developer who just wanted a smarter home. Engineered by Ekimtech