GTC 2026 | Workshop 09：LLMs & On-Device AI Architecture

​💡 Workshop focus on:
When inference cost becomes a ceiling for growth, your architecture is your competitive advantage. This session focuses on strategies for cloud and on-device model co-ordination: when to run on-device, how to split the workload, and techniques like caching, compression, and routing to balance experience, privacy, and cost.

​Workshop format: 1-to-10 enablement session

​🚀 What You’ll Learn:

• ​Understand the capabilities and trade-offs of cloud vs. on-device AI and common architectural patterns

• ​Master the key levers for inference efficiency: latency, throughput, cost, and energy consumption

• ​Design layered invocation strategies: small-model-first, routing, fallback, and caching

​🤖 Who It’s For:

• ​AI engineering leads, architects, and product/tech owners

• ​Teams building agents, developer tools, mobile AI, or smart hardware

• ​Projects hitting a wall with inference cost or latency