Freddy L. Email and Phone Number

• I led the project to automate the team release process, from artifacts building, and release, to seamless deployment. This has significantly increased work efficiency. We are 16x faster on artifacts building from 8 hours to within 30 mins. We are in automation mode compared to the heavy engineer intervention one that rewards us with Modeled Change Management(MCM) free during production release.• I took a leading role in driving many areas of supporting the EMR Serverless Debug Jobs feature.

• I drove the end-to-end design and implementation with cross-team collaboration to simplify the team AMI release process that helps reduce operational costs by 200%. It plays a key role in the team to scale and helps move fast to bring more value to customers when supporting EMR debugging experience in EMR Serverless.• As a major contributor, I delivered the Amazon EMR Studio SQL Explorer feature. It brings the next level of user experience for Data Scientists and Data Engineers to browse the data catalog and run SQL queries on EMR clusters from EMR Studio.• I, as a major contributor to the Warm Pool Automation project, help our team to leap toward the goal of a full CI/CD AMI release process. The project helps my team achieve the following savings: 1) Reduce release executor effort for MCM execution by 500%, 2) Reduce calendar time for MCM execution by 200%, 3) Reduce the MCM execution steps by 1500%, 4) Reduce various release dependencies

EMR customers only had access to YARN Application Timeline Server (ATS) on running clusters by setting up a web proxy through an SSH connection. To resolve the pain point, this project aims to provide customers with a better debugging tool by making ATS available for terminated clusters (running off the cluster). • I worked on an open-source Hadoop YARN project, ironed out existing code logic, and modified it to log data into files in HDFS on a daily basis in addition to LevelDBs. This resolves the difficulty of maintaining a consistent copy of LevelDBs off the cluster to be accessed while the cluster is terminated. • I implemented the log-removal feature so that old logs stored in HDFS existing longer than a certain time can be removed to save HDFS space. Both the writing and log-removing features may be contributed to the open-source community later at a proper time to reduce maintenance overhead and rebase efforts in the future.• I implemented a reader so that off-cluster ATS can read from a designated AWS S3 bucket and reconstruct the LevelDBs for off-cluster ATS.• I upgraded EMR default YARN Application Timeline Service from version 1.0 to 1.5, which will be included in EMR release 5.30, and I also set up a testing procedure for upgrading the YARN timeline service.

Product Ops Readiness Tasks • I leveraged AWS CloudFormation to build Amazon CloudWatch dashboard for important necessary metrics to monitor product service healthiness• I set up CloudWatch alarms to send out notifications when triggered and integrated them into Amazon Carnaval Alarms to cut tickets to related team• I wrote end-to-end Integration Tests for product APIs and deployed them to the production environment to ensure functionality whenever code changes• I implemented Canary Tests for key part service to conduct minimal testing on a continuous basis