In modern AI-driven platforms, latency is not just a performance metric—it directly translates to revenue gain or loss. This talk explores how Service Level Objectives (SLOs) should be redefined when systems sit on the critical path of monetization, such as pricing engines, personalization models, and real-time decision systems. Drawing from real-world experience building large-scale AI platforms at companies like GoodRx and Meta, I will share failure and recovery stories where milliseconds impacted millions in revenue. We will break down how to design SLOs that align with business metrics (not just uptime), how to detect hidden degradation before revenue drops, and how to build resilient recovery loops across ML and distributed systems.
Attendees will walk away with practical frameworks for:
Chenghao Liu is a senior AI engineer, entrepreneur, and founder of Aivy, an AI infrastructure and customization company backed by the UCLA Venture Accelerator. She has led large-scale AI and distributed systems across leading technology companies including Meta, Amazon, Microsoft, and GoodRx. At GoodRx, she led AI-powered personalization and pricing systems that supported billions of transactions and contributed to hundreds of millions in revenue impact. Previously, at Meta, she worked on AI infrastructure and monetization systems within the metaverse ecosystem.
Chenghao holds a Bachelor’s degree in Computer Science from the University of Illinois Urbana-Champaign and is currently pursuing an Executive MBA at UCLA Anderson School of Management. Her work focuses on building production-grade AI systems where performance, reliability, and business outcomes are tightly coupled.
