In multitenant AI clusters, bandwidth is not the SLA—Job Completion Time (JCT) is. In this joint session, NextHop.ai and Keysight present a real-world study of how load balancing modes, ECN thresholds, and congestion control tuning directly impact JCT in shared AI Ethernet fabrics. Using production-class switching and traffic emulation, we demonstrate how overly conservative settings can double completion times and waste GPU cycles—and how systematic tuning makes JCT predictable. Attendees will gain a practical framework for designing and validating AI networks around measurable performance outcomes, not just throughput.
Jess Tarkar is a Solutions Architect at Keysight Technologies. She is an accomplished technology professional with over a decade and half of deep expertise in AI-driven infrastructure, network validation, wireless systems, and test solutions. She builds strategic partnerships with hyperscalers, neocloud providers, enterprises, and partners to translate emerging AI infrastructure challenges into practical validation and benchmarking strategies. She has a proven track record of driving innovation and delivering customer-centric solutions that enable customers to architect, validate, and optimize high-performance, next-generation AI networks at scale.
Juuso Lehtinen is a networking engineer with over 20 years of experience across AI, data center, and service provider networks. He is currently a Proof of Concept Engineer at Nexthop AI, focusing on validating AI-driven networking architectures and supporting customer deployments. He previously spent 12 years at Arista Networks as a Principal Systems Engineer, providing technical leadership to hyperscale and tier-2 cloud customers building large-scale data center networks. Earlier in his career, he spent seven years at Tellabs in multiple engineering roles supporting IP/MPLS edge routing platforms through system test, product launch, and customer-facing proof-of-concept engagements.