At the Spark Summit in Brussels, MediaMath’s SVP of Data Science, Prasad Chalasani, gave an invited keynote talk, Extreme Scale Ad-Tech at MediaMath with Spark and Databricks. MediaMath’s demand-side platform responds to over 200 billion ad-opportunities daily, and leverages massive amounts of data to power smarter digital marketing. We use Spark heavily both in production and R&D to develop innovative, proprietary, and scalable solutions to multiple large-scale data problems, such as: Training Machine-learning models for predicting conversion probability given an ad-impression Measuring causal effectiveness of advertising using randomized tests Estimating audience reach for specified targeting criteria. Finding deviceIDs belonging to the same user based on […]
In a data-streaming web world, things happen fast. In less than the blink of an eye, MediaMath’s digital marketing systems host real-time auctions and serve ads across the world to the tune of 4.6 million queries per second. In this session at the GOTO Chicago Conference, MediaMath CTO Wilfried Schobeiri dove into MediaMath’s data stream processing architecture and how the company is building the next generation of real-time, high performance systems in Go. Using Go, MediaMath is able to scale its systems on a minimal resource footprint. Wil also explains why Go is a game-changer for building services, how to […]
In this blog post, I will share a proof of concept for real-time attribution using Apache Flink from streaming data sources of impressions and events, and how we handled some of the specific problems inherent in windowing and processing real-time data streams at scale. Our goal was to determine if we could use Flink to stream impression and event data so that we could determine attribution in real time in order to optimize advertising strategies immediately. In digital advertising, we refer to ads – whether they are served on social networks, Mobile, Video, or display – as impressions. Once the […]
Owein Reese, who manages a group of engineers in MediaMath’s Creatives Management Tribe, took an unusual route to a career in software development. Owein’s mother was a programmer at IBM back in the days of punch-card programming (well… the 80’s) but punch cards failed to interest him in programming as a child. It wasn’t until he got to college that he began to write code to solve mathematical problems and still later on in his career that software became the focus. Owein entered the University of Rochester with degrees in optical engineering and math at the peak of the dotcom […]
Last week, MediaMath employees around the world tuned in for our bi-annual hackathon, and hosted more than a dozen teams working for 24 hours to build projects that wouldn’t otherwise make out sprint schedule. Joe Zawadzki, our CEO, and Byron Ellis, the CTO of Spongecell and our guest judge, shared their thoughts immediately after watching presentations from the participants.
Software design pattern is a general repeatable solution to a commonly occurring problem in software design. It provides a description and guideline to solve a problem that can be used in multiple different situations. Because development speed is increased when using a proven prototype, developers using design pattern templates can improve coding efficiency and final product readability. MediaMath’s Engineering team used design patterns to add flexibility, extensibility and reusability to components of a greenfield real-time sizing service for Data Management Platform (DMP). Advertisers use a DMP to store millions of data entries that they have on potential users they would like to […]