Building AI Applications on Distributed Systems

We're excited to welcome Madhushree Kumari, WomenTech Network Founding Member, as our featured guest speaker!

As AI applications continue to grow in scale and complexity, building systems that can efficiently train, deploy, and serve large models has become a critical engineering challenge. Modern AI workloads, particularly large language models and deep learning applications, demand distributed architectures that can deliver high performance, resilience, and cost efficiency across cloud environments.

In Building AI Applications on Distributed Systems, Madhushree Kumari will explore the architectural principles behind scalable AI infrastructure and the techniques engineering teams use to support demanding AI workloads in production.

This session will examine how distributed systems enable modern AI applications by covering topics including:

• Scaling AI workloads using data parallelism and model or pipeline parallelism across distributed clusters.
• Managing GPU resources and coordinating distributed compute for efficient training and inference.
• Reducing the impact of network latency and optimizing parameter synchronization across nodes.
• Building resilient AI systems through checkpointing and fault-tolerant design.
• Improving throughput while balancing infrastructure performance and cloud cost efficiency.

Drawing on more than 13 years of experience designing large-scale distributed systems, Madhushree will share practical architectural approaches that help engineering teams build AI platforms capable of supporting modern production workloads.

Whether you're developing AI-powered products, designing distributed infrastructure, or expanding your knowledge of cloud-native AI architectures, this session will provide practical insights you can apply to future AI projects.