Local LLM lab
Self-hosted inference environment on Proxmox for testing open models, prompt workflows, and offline agent behavior without relying on external APIs.
Active
AI Lab
Experiments in automation, LLM tooling, and applied AI.
A working space for AI-adjacent projects: agent workflows, MCP integrations, monitoring automation, and the tooling that ties them into real infrastructure. Practical, self-hosted, and built to run in restricted environments.
Projects
Active work
MCP tooling
Exploring Model Context Protocol for connecting LLM workflows to internal systems: file shares, SQL Server, monitoring endpoints, and task schedulers.
In progressAutomation monitoring
Python-based monitoring with logging, recovery logic, and Windows notification hooks. Extending it with anomaly detection and AI-assisted log summarization.
Activen8n AI workflows
Orchestrating document processing, notifications, and data pipelines with n8n, integrating LLM steps where they add real value over deterministic logic.
ActiveRTL web tooling
Arabic-first internal web apps with AI-assisted form validation, summarization, and search. Applied to real business workflows backed by SQL Server.
PausedDistributed systems notes
Reading and experimenting with consensus, replication, and scheduling. Some of this feeds back into the automation and monitoring work above.
OngoingNotes
Recent write-ups
Stack
Tools I actually use
Models & Inference
- Ollama
- llama.cpp
- Open WebUI
Tooling & Protocols
- MCP
- n8n
- Playwright
Infrastructure
- Proxmox VE
- Docker
- Ubuntu Server
Languages
- Python
- TypeScript
- SQL