San Diego, California, United States
Cloud task scheduler
• Developed a company-wide task scheduler in Python to perform cloud computing tasks on Amazon EC2 Instances, handling up to ~6,000 tasks per second and 4,000 instances through EC2 spot fleet.
• Minimized workload cost by implementing a task scheduling algorithm to dynamically adjust instance size and instance computing power, based on task count and resources needed per task.
• Decreased ~2.5% instance utilization by diminishing redundant instance initialization, reducing ~$15k budget monthly.
Internal tool
• Implemented a Python and SQL based solution to optimize the database scripting efficiency for updating mass data.
• Developed a wrapper API in Python to batch transactions concurrently for database scripting, increasing speed by 4x and reducing 80%+ memory use.
• Logged data changes and transaction events using PG/plSQL, and developed a log-based rollback function to assist with batch scripting; visualized logs in a tabular view by developing 4 web pages in Vue.js.