Speaker
Description
INK (Interactive aNalysis worKbench) is a web-based platform that opens the compute and storage resources of batch computing clusters to physicists — enabling interactive data analysis on demand, anytime and anywhere, with just a browser and no local environment setup. In this talk we present the recent progress of INK relevant to LHAASO. We introduce the container-native one-click deployment, which allows a small site or group to set up a production-ready analysis environment on a few machines with a single command. We describe how INK is customized for individual experiments — for example, dedicated GPU resources and applications that are visible only to LHAASO members. We present JobLens, our eBPF-based job-level observability service that gives users real-time, per-job views of CPU, memory, I/O, network and GPU usage. Finally, we show how AI agents (such as OpenCode/OpenChamber) are integrated as first-class interactive applications, backed by the IHEP AI platform, so that researchers can use AI assistance out of the box.