2nd Annual AI Infra Innovation Night - Inference Optimization
From Compute & Memory to Cost per Successful Task
Back by popular demand, our next AI Infra Night dives into one of the most important challenges in AI infrastructure: inference optimization.
As AI moves into production, the challenge is shifting from simply accessing compute to getting more intelligence from every infrastructure dollar — across model execution, serving infrastructure, routing and agent runtime.
Co-hosted with CANOPY, a Stanford student association, the evening will bring together AI researchers, infrastructure engineers, technical founders, enterprise AI leaders and investors for in-depth technical discussions and networking.
Event Details
Date: Tuesday, November 17, 2026
Time: 4:30 – 8:30 PM
Location: Palo Alto, CA — Exact venue details shared upon registration
Food and beverages will be provided.
What to Expect
One keynote and three expert panels exploring inference optimization across the stack:
Optimize the Machine → Optimize the Serving Stack → Optimize the Runtime
Tokens/$ → Cost/Request → Cost/Successful Task
From making individual model executions more efficient, to serving models at scale, to optimizing the cost and reliability of completing real-world AI tasks.
Agenda
4:30 – 5:00 PM — Arrival, Check-in & Networking
5:00 – 5:15 PM — Opening + Keynote from Head of AI, Confluent
5:15 – 7:00 PM — Expert Panel Discussions and Q&A
Panel 1 — Optimize the Machine
Model Execution, Compilers, Kernels & Hardware-Aware Optimization
How can advances in model architecture, compilation, kernels and hardware-aware execution deliver more intelligence from the same compute budget?
Panel 2 — Optimize the Serving Stack
Serving, Scheduling, Caching & Infrastructure Utilization
How do we efficiently serve models at scale through better scheduling, batching, KV-cache management, workload placement and infrastructure utilization?
Panel 3 — Optimize the Runtime
Routing, Context, Agent Execution & Adaptive Optimization
How do we dynamically orchestrate models, context, tools and agent execution to optimize cost, quality and reliability across the entire task?
7:00 – 8:30 PM — Drinks, Food & Curated Networking
Speakers & Attendees
Speakers will include founders and technical leaders from emerging inference optimization companies such as TensorMesh, Subconscious.dev, Inco.ai, NeuroSpark, Eigen AI (acquired by Nebius) and Superlinked, alongside senior AI leaders and practitioners from Oracle, TikTok and Hippocratic AI, and investors from Intel Capital, Samsung, Alumni Ventures, Mayfield, Touring Capital, M12, Sierra Ventures and Laude Ventures.
The evening will bring together approximately 100–120 curated attendees, including AI researchers, infrastructure engineers, founders, senior AI practitioners and 30–40 AI infrastructure-focused investors, for technical discussions and meaningful connections across the AI infrastructure ecosystem.