This keynote explores about the Goodput which is reframing the LLM training efficiency from raw tokens/sec to real progress. The talk decomposes the losses across infrastructure availability, framework checkpoint/recovery overhead, and model compute utilization (MFU) and yields an end-to-end efficiency metric that guides actionable engineering priorities.
Enterprise Data Architect
Cognizant Technology Solutions
Platform & Devops Enablement Manager | Reliability & Release Engineering
Comerica Bank, USA
Sr. SAP Architect
CEO
TTNC Teknoloji
Manager
QNB Insurance
CEO
Relevant Box LLC
Sr. Software Engineer
Tata Consultancy Services
Software Development Engineer
Capital One Financial
Senior Software Engineer III,
Softility Inc.
Principal Engineer,
Amazon Web Service
Business Analyst / Product Owner,
Verifone
Lead SAP, TractorSupply Company,
Brentwood, TN, USA
Solution Architect | AI Researcher | Ieee Senior Member| IETE Fellow | Cognizant Technology Solutions, USA
Senior software engineer,
MeijerCloud Infrastructure (OCI)
Oracle
Principal Architect,
Hakkoda IBM
IT Teacher,
Washington University Of Science And Technology