Local LLM – Single User vs Multiple Users
How to serve a local LLM on GPU with llama.cpp for yourself, or with vllm when you need to share it with a team.
Blog
A few notes from experiments, research engineering, and side projects.
2026
Distilling a Winning Ticket with SymTorch
Exploring symbolic approximation of winning tickets in neural networks using SymTorch
Ephemeral conda enviroments
I bet you’re also tired of cloning repos online that use conda as package manager