Presentation slides available here!

FOUNT co-PI and Chameleon user Professor Fraida Fund (NYU) presented her full curriculum and FOUNT courselets used in her Machine Learning Operations (MLOps) graduate-level class in Spring 2025. Prof. Fund managed to design and implement a fully interactive course in which over 190 students deployed ML services using popular Infrastructure-as-Code (IaC), i.e., Terraform and Ansible, trained and optimized ML models for deployment in various environments (virtual machines, bare metal CPUs and GPUs, edge devices), and configured real software and hardware infrastructure on Chameleon’s resources to support a fully operational ML system. The class consumed significant resources on Chameleon, including over 180,000 instance hours on bare metal and VMs, ~11TB of block and object storage, and over 50,000 floating IP hours, which – by conservative estimation – would have cost around $50,000 in total to host on commercial cloud services (AWS or Google Cloud). Read the full paper.

Prof. Fund didn’t just teach the class; she also packaged her materials for public release on Trovi and produced an instructor guide for folks interested in reusing or adapting her course for future students.

Check out all the materials. They are worth it!