Ksheer Sagar Agrawal

I build systems for machine learning at scale. Previously optimized distributed caching at AWS DynamoDB; now extending that to training, inference, and agentic workloads.

I approach ML from a systems perspective. How do you train models efficiently? How do you serve them with low latency? What infrastructure breaks when you scale? These questions matter to me more than model architecture alone. At AWS DynamoDB, I learned that systems reliability is unglamorous but critical—p99.9 uptime, fast releases, knowing where your bottlenecks are. I’m extending that mindset to ML infrastructure.