Who We Are
Does advertising work on Twitter? When an advertiser runs an ad campaign, they are trying to generate some impact on their business, whether increasing brand awareness, or driving users to visit websites, install apps, or make purchases. We provide data to prove the effectiveness of ads campaigns. It’s critical for our advertisers to understand the influence of each advertising impression on a consumer’s awareness, intent, and purchase decision. We are building a highly accurate real-time streaming system with the largest scale of data at Twitter to make this vital information available to internal engineering teams, and to our advertisers.
The timely and accurate view we provide of the real-world impact of our ads feeds into a number of critical functions inside the Revenue organization. The most obvious of these is that the record of the value we’re generating flows into our advertiser-facing surfaces and APIs, enabling our customers to make informed decisions about how to allocate their budgets and optimize their campaigns. The data also flows into our ML prediction models, ensuring that our ad delivery system is learning and optimizing based on the latest available data. Finally, this data is available for analysis and inspection by our product data scientists, allowing our product managers to make data-driven decisions around which features to prioritize. We are a critical part of making ads successful at Twitter.
We primarily code in Scala and our tech stack is comprised of Finatra servers running Kafka Streams to process and join streams of hundreds of thousands of Kafka messages per second, with additional storage in RocksDB, Memcached, GCP Bigtable, and HDFS storing hundreds of terabytes. We also do offline data processing in Druid and Hadoop with Scala.
What You’ll Do
You will work on all aspects of the attribution pipeline to ensure low latency delivery of accurate attribution data to Twitter's advertisers and internal systems.
Work with our product managers to understand the requirements for new attribution products and features, design and implement the new APIs and changes to the pipeline.
Partner with platform engineers to help scale our systems to handle additional use cases and Twitter’s organic growth.
Interface with one of our machine learning teams to enhance the prediction feedback loop with additional attribution signals.
Observe and exploit the weak points of our system and build guardrails to increase the fault tolerance, leveraging features in our open source platforms and contributing back to those projects where appropriate./
Combine a deep understanding of our product requirements with system knowledge to perform storage and processing optimizations for cost savings.