Systems-Facing Track – September 2026

What is Aleph?
Designed and built by the Research Computing team at the University of Alberta – the folks supporting the computational infrastructure used by the Alberta Machine Intelligence Institute (Amii), Aleph is a complete multi-model inference platform built to serve the full breadth of research science from a single endpoint. It combines a self-deploying RKE2 Kubernetes cluster, HAMi GPU virtualization, KServe/Knative serving, more than 170 model definitions, and a FastAPI gateway that unifies them all behind one multi-LLM-compatible API. Protein structure, genomics, materials, climate, astronomy, medical imaging, and large language models all share one key, one URL, and scale-to-zero economics.

Aleph is not a chatbot stack. CERN runs a similar KServe-based platform for physics inference at ml.cern.ch, and Aleph is that idea applied across all of research science. The session walks through the architecture and the thinking behind it, shaped by the DOE’s Genesis Mission push toward autonomous science, DeepMind’s agent work, the wave of Open-Weight models, and CERN’s production platform.

Speakers:
Rahim Khoja, HPC & AI Infrastructure Engineer, University of Alberta
Karim Ali, HPC & AI Infrastructure Engineer, University of Alberta

Connection info can be found on our Slack channel or through the mailing list