LLM examples

This repository contains some examples of LLM deployments in Aalto University premises.

Use local models on Triton

Huggingface transformers

See this document.

Batch inference via vLLM

See this document.

Server via vLLM

See this document.

Use Aalto's open-source LLM API

See this document.

Agent configurations

See this document for configuration examples for AI coding agents using Aalto's Azure and LLM Gateway endpoints.

Agent skills for Triton

Task-workflow skills for AI coding agents on Triton live in a separate repository: AaltoSciComp/triton-skills.

SciComp Docs MCP

See this document for connecting AI coding agents to the SciComp documentation search MCP server, so they can look up live Triton docs instead of guessing.