ASSEM BEKBOLATKYZY
Seattle,WA
Overview
2
2
years of professional experience
Work History
Software Engineer
Amazon
Bellevue, USA
06.2025 - Current
- Developed and implemented an idle pod termination system that monitored GPU utilization metrics, automatically terminating pods with less than 5% GPU utilization over a three-hour period, leading to a reduction of approximately 1000% in GPU costs daily and the termination of around 5,000 idle pods weekly.
- Leveraged AWS Lambda and Kubernetes automation to ensure safe pod termination, overcoming challenges such as maintaining cluster stability and preventing disruption, resulting in a scalable solution that managed approximately 15,000 p5 GPU nodes and significantly improved resource utilization.
- Implemented comprehensive workload tagging for Kubernetes jobs and applications, enhancing cost allocation and ownership tracking, and achieving an approximate 90% improvement in reporting accuracy and resource visibility at a scale of 6000 pods submitted daily.
- Developed and implemented a GPU utilization reporting pipeline by integrating data from AWS CloudWatch, Athena, Prometheus, and Grafana, resulting in a reduction of idle GPU time by 40% and achieving cost savings of approximately $1.2M monthly.
- Leveraged AWS Glue, Lambda, and internal AI frameworks to automate data collection and reporting, enabling scientists and leaders to make data-driven decisions and improve GPU resource usage, leading to an 2x increase in utilization score and reducing management's manual reporting effort by 50 hrs monthly.
Software Engineer Intern
MongoDB
San Francisco, USA
06.2024 - 08.2024
- Successfully migrated 75% of legacy Backbone.js components to React in MongoDB's Atlas Clusters via Java and JavaScript, resulting in a 30% improvement in cluster performance and reducing maintenance time by 40%.
- Designed and implemented a full-stack key vault system with TSX and Java, integrating with Azure Key Vault's private endpoints. Leveraged Azure services and API integrations to ensure secure and efficient CRUD operations, enhancing data security, resulting in a 25% increase in operational efficiency.
- Conducted comprehensive unit, integration, and end-to-end tests using JUnit, Mockito, and Cucumber on all components, achieving 98% test coverage. Utilized Bazel for development and implemented continuous integration, reducing bug-related incidents in Evergreen by 35%.
Education
Bachelor of Science - Computer Science and Finance Major
Minerva University
San Francisco, CA05-2025
Skills
- Python, Java, JavaScript and R
- Machine learning and NLP
- PostgreSQL, MySQL, MongoDB
- Django and Flask
- Unittest, JUnit, Mockito
- Containerization with Docker, CI
- Google Cloud, Azure, AWS
- Kubernetes, Linux, Unix
Timeline
Software Engineer
Amazon
06.2025 - Current
Software Engineer Intern
MongoDB
06.2024 - 08.2024
Bachelor of Science - Computer Science and Finance Major
Minerva University