LLM examples
A set of examples for deploying LLMs on Aalto University's Triton cluster, covering local inference through Hugging Face Transformers and vLLM, Aalto's own LLM API, and configuring AI coding agents against Aalto's endpoints.
This repository contains examples of LLM deployments on Aalto University's compute premises, covering running local models on the Triton cluster through Hugging Face Transformers and vLLM, both for batch inference and as a server, and using Aalto's open-source LLM API.
It also includes configuration examples for pointing AI coding agents at Aalto's Azure and LLM Gateway endpoints, and for connecting agents to a SciComp documentation MCP server so they can look up live Triton docs instead of guessing. Separate task-workflow skills for AI coding agents on Triton live in the AaltoSciComp/triton-skills repository.
Written in Python. MIT licensed.