Amaresh Bingumalla will showcase how his team revolutionized their enterprise data infrastructure by adopting open-source technologies, achieving over $1M in savings each year. They leveraged Apache Hudi to build a scalable data lake with incremental processing, providing engineering and machine learning teams with seamless access to data for analysis and model training. By replacing costly SaaS solutions like Splunk and Segment with a custom observability stack built on OpenSearch and an in-house event processing platform, they eliminated expensive vendor dependencies without compromising system reliability. This session will provide practical insights into evaluating open-source alternatives, building scalable self-service data workflows, and measuring the ROI from transitioning from SaaS to custom-built solutions.