A fast-growing AI SaaS product

A fast-growing AI SaaS product needed to scale its model inference infrastructure to handle enterprise-level traffic without downtime or latency spikes. I architected and deployed 18 microservices on AWS handling 50M+ daily API requests, serving 25,000+ concurrent users. Migrated a legacy monolithic system to a cloud-native architecture, reducing infrastructure costs by 42% and improving system reliability from 95% to 99.97%. Also designed the real-time ML inference pipeline that achieved a 78% latency improvement and supports 15+ production models in deployment.