Amazon

SDE Intern | Seattle, WA | Summer 2019

In Summer of 2019, after my junior year of college, I was a SDE Intern at Amazon. I was placed in their headquarters in Seattle within the Amazon Advertising organization. Specifically, my team supported the back-end system behind Video in Search. This project was done under the guidance of Irakli Zhuzhunashvili and Braiden Judd.

Project Summary

My overall goal for the summer was to allow my team to better understand the end-to-end ad campaign creation. In response, I came up with two projects that helped us measure how many and why ads were being “lost” during the creation process.

Video Content Failure Metrics

There are many ways that videos uploaded to become ad content may fail technically- such as being too long or not having the correct aspect ratio. My first project logged data on why ads failed this phase of validation and created visualizations of these data through an Elasticsearch Kibana dashboard. This allowed team members to see trends over time with both the number and cause of failures.

Timing Ad Moderation

One major pain point in the ad campaign creation process is moderation. It can sometimes take much longer than desired or expected for an ad to get moderated. We wanted a way to measure how long ads spent in moderation. I leveraged an Amazon Advertising internal tool which continually published updates to the central ad database that can be filtered and data extracted to be processed. This allowed us to detect when an ad is created and moderated, along with the moderation result, to produce metrics on timing.

Skills Gained & Used

Kotlin
AWS tools (including Lambda, S3, CloudWatch, and Dynamo DB)
Elasticsearch & Kibana
Design Documents

Deliverables

The metric gathering and dashboards created from my project are in production and used by my team (Available Internally Only)

“Thanks for the time you invested in setting up telemetry measurements for the UCB advertiser experience end to end. I’m eager to use your Kibana dashboards to track completion rates and see if/how we can apply what you’ve built for [other ad products] as well. I also appreciate the time you invested in improving the intern experience for the rest of the team through planning the fulfillment center tour and suggesting the fireside chats.”