Harry Levinson Email and Phone Number

● Enabled 10+ internal teams to better manage 30 million Equipment Service data entries by backfilling, reconciling, and adding automated validation of attribute Equipment Class Service schema data sets.● Designed and implemented a shipment event processing workflow using AWS lambda, SNS, and SQS to provide product managers and data center technicians with the most accurate patch rack landing dates in 13 AWS data center clusters.● Built workflows synchronizing DynamoDB and RDS tables with a data lake repository to enhance business teams' access to cabling and reservation status data from 13+ AWS data center clusters.● Automated legacy brick hostname reservation logic covering 30+ data center sites to eliminate requirement for on-going manual data correction. ● Wrote thorough unit tests with at least 80% code coverage of new business logic, and integration tests.● Added metrics, alarms, monitors, and logs to critical services, and solved high-impact customer issues.● Mentored intern to build API via Elastic Search by reviewing code, testing methods, and API and system design.

• Built a model training service to construct models that estimate server space and power supply resource demand based on forecast server demand via Python, an AWS step function, and an API Gateway interface. • Designed a model training system to assess out-of-sample accuracy, export models to JSON configuration files in DynamoDB for production, and produce visualizations to illustrate model performance. • Constructed baseline models to translate forecast server demand into estimated future power and space demand in AWS data centers.

● Assessed accuracy of newest version of ASPECT, an open-source thermal convection simulation software package, using the Simple Annulus, SolCx, and SolKz benchmarks (I re-computed last summer’s results to include in a paper).● Worked extensively on the UC Davis supercomputing cluster Ymir and submitted batch jobs using Slurm. ● Coded shell and Python scripts to automate the collection, organization, and analysis of velocity and pressure error data. ● Performed some minor development work for ASPECT in C++. ● The research was approved and funded by the National Science Foundation.

● Collaborated in a team of six to test the performance of ASPECT. Analyzed the accuracy of ASPECT using different particle interpolation schemes on many different benchmarks, including the SolCx and Burstedde benchmarks. ● The research was approved and funded by the National Science Foundation.

Utilized the CodeIgniter PHP Framework to co-develop a Parent-Teacher Conference Scheduling program with the Kentfield School District’s Technology Director.

● Co-developed a web application in Python to manage multiple databases of student and staff information with the Kentfield School District’s Technology Director and Technology Assistant. ● Performed unit testing to validate the code efficacy.● Migrated teacher websites to a new content management system.