01
Local LLM Benchmark Lab
A reusable Windows PowerShell benchmark for llama.cpp. It captures model throughput, timing, GPU and VRAM behavior, CPU load, RAM use, and structured CSV results across repeatable tests.
Projects
Experiments with evidence.
These projects are part build log, part learning system: practical work with enough measurement and documentation to make the results useful again.
02
Home Server Architecture
A power-conscious home-server design for redundant storage, virtual machines, secure remote access, and room for a moderate GPU as the lab grows.
03
Data Engineering Notes
A developing set of explanations and reusable patterns covering Databricks, PySpark, Azure, Microsoft Fabric, and the architecture decisions behind dependable data workflows.