ICICLE Services Board
Cyberinfrastructure services for automated data labeling, image segmentation, model serving, and edge deployment, usable across any field.
ICICLE (Intelligent Cyberinfrastructure with Computational Learning in the Environment) is a national NSF-funded AI institute focused on making powerful machine-learning and computing tools accessible to researchers who aren't infrastructure experts. Its offerings are organized into two tiers. The domain-agnostic foundation services work across any field and include out-of-distribution detection (flagging data that doesn't match what a model was trained on), smart labeling and zero-shot image segmentation (generating training labels without a hand-labeled dataset), a no-code image-processing lab, Patra model cards (documenting a model's provenance, performance, and fairness), a flexible multi-model server for serving many ML models through one interface, HPC job-submission optimization, and intelligent orchestration for deploying and managing AI across edge devices. On top of these, the domain-specific services apply the same infrastructure to three real-world areas: digital agriculture, animal ecology, and food logistics and security. Everything is browsable on a single Services Board, runs on ICICLE's underlying Tapis framework, and the code is openly available on GitHub.
All services are free to access. You just need to sign up for a free TACC (Texas Advanced Computing Center) account to log in and use them.
