Scaling Agentic Inference Across Heterogeneous Compute with Zain Asgar - #757
In this episode, Zain Asgar, co-founder and CEO of Gimlet Labs, joins us to discuss the heterogeneous AI inference across diverse hardware. Zain argues that the current industry standard of running all AI workloads on high-end GPUs is unsustainable for agents, which consume significantly more tokens than traditional LLM applications. We explore Gimlet’s approach to heterogeneous inference, which involves disaggregating workloads across a mix of hardware—from H100s to older GPUs and CPUs—to optimize unit economics without sacrificing performance. We dive into their "three-layer cake" architecture: workload disaggregation, a compilation layer that maps models to specific hardware targets, and a novel system that uses LLMs to autonomously rewrite and optimize compute kernels. Finally, we discuss the complexities of networking in heterogeneous environments, the trade-offs between numerical precision and application accuracy, and the future of hardware-aware scheduling.
The complete show notes for this episode can be found at https://twimlai.com/go/757.
Pixie, with Zain Asgar and Ishan Mukherjee
Pixie Labs built an observabiity platform for Kubernetes, which uses eBPF to get telemetry without user intervention. They were recently acquired by New Relic, who open sourced the Pixie software. Co-founders Zain Asgar and Ishan Mukherjee join Craig Box to tell the story and talk about what's next. Guest host Alex Ellis tends his garden.
Do you have something cool to share? Some questions? Let us know:
web: kubernetespodcast.com
mail: kubernetespodcast@google.com
twitter: @kubernetespod
Chatter of the week Episode 116, with Alex Ellis
GrowLab Announcement blog
Alex's talk at the GIFEE Day
Monty Don
OpenFaaS in the RISC-V keynote
New Kubernetes on Edge training course
News of the week eBPF for Windows
GKE Dataplane V2 is GA
Confluent for Kubernetes GA
VMware Tanzu SQL, with MySQL, for Kubernetes, 1.0
VMware Modern Apps Connectivity Solution
Do the State of DevOps survey!
Links from the interview Pixie Labs
What is Pixie overview slides presented to CNCF
Public beta launch and announcement of Series A funding TechCrunch coverage
Pixie Labs acquired by New Relic; New Relic acquires Pixie Labs
A day in the life of a Kiva robot
Recognition for Google Lens clothing recognition
Dog or blueberry muffin?
Episode 125, with Ramiro Berrelleza
How Pixie Works
New Relic goes all-in on OpenTelemetry and Open Source
Pixie on GitHub
Pixienauts community
New Relic upgrades to Platinum member at CNCF
Zain Asgar and Ishan Mukherjee on Twitter