Sabha Parameswaran

EC2 Image Builder pipelines sprawl fast, and frontier agents that run for hours unattended need guardrails. In this builders' session, decompose Image Builder operations into reusable agent skills run via Kiro, Claude Code.

Mastercard moved beyond the "Multi-Catalog" struggle and successfully federated Databricks Unity Catalog to AWS Glue. This allows Apache Spark engines to consume data in Apache Iceberg tables that register in Databricks Unity Catalog and sync seamlessly across AWS services like Amazon EMR, enabling ACID reliability and multi-engine interoperability without vendor lock-in. Mastercard achieved a Single Source of Truth for governance allowing engineers to use the best engine for the job without metadata fragmentation. Mastercard achieved seamless catalog federation, allowing domain teams to choose optimal compute engines. This open-format strategy created a sustainable, future-proof architecture supporting machine learning, streaming, and governed data sharing across the enterprise.

Organizations that have migrated to the cloud via lift and shift or that are already on the cloud with a legacy architecture need to refactor and modernize their environment to fully realize the benefits of the cloud (e.g., agility, cost, and economy of scale).

As a result of digital transformation, many organizations have adopted a hybrid approach, using both public cloud and on-premises data centers. Public cloud is used primarily for modern microservices, processing big data, IoT, and AI/ML workloads, while on-premises servers continue to run legacy applications that drive manufacturing, require low latency for real-time response like bidding,...

Is your organization currently using traditional on-premises applications and middleware stacks but looking to move to the cloud? In this chalk talk, review steps for modernization along with a high-level vision of how to translate, refactor, and rearchitect for the cloud.