Industrial Panel - SIGMETRICS 2025

Industrial Panel: SIGMETRICS whitebox wisdom in an AI blackbox era

Date: Wednesday, June 11, 2025

Time: 4:00 pm ~ 5:30 pm

Venue: Wang Center Theater

Panelists:
	Carole-Jean Wu (Director of AI Research, Meta)
	Siddhartha Sen (Principal Researcher Manager, AI Frontiers Lab, Microsoft Research)
	Ramesh K. Sitaraman (Chief Consulting Scientist, Akamai)
Moderator:	Y.C. Tay (National University of Singapore)

Synopsis:

The panel will give an industry perspective on

(1) How will the rise and rise of blackbox ML models impact SIGMETRICS?
(2) How can SIGMETRICS contribute to the performance analysis of ML systems?
(3) How can SIGMETRICS ride on this rise of ML models?
(4) How can SIGMETRICS connect with the ML community and industry?

Brainstorm:

The panelists will flesh out some thoughts from their pre-conference discussion:

Multi-cluster/multi-GPU training/inference optimization. Environmental impact and sustainability. MLOps (cf. DevOps). Resource constraints: distilled models and small LMs. LLM inference and serving performance. Whiteboxes for LLM agents, RAG (Retrieval-Augmented Generation) and LAMs (Large Action Models). Metrics, benchmarks, and LLM-as-a-judge. Modeling the model. Internships and jobs.

Time will be allocated for questions and contribution from the audience.