Recommendation System for AdTech Company

In the competitive landscape of adtech, precision in auctioning is paramount for securing optimal deals for publishers. The pursuit of sophisticated solutions is key to maximizing outcomes.

The Challenge

The challenge faced by the Israeli adtech company stemmed from the obsolescence of their tech stack and machine learning algorithms, particularly evident in their use of XGBoost, which proved inadequate for handling categorical variables like ad content. Thus, the imperative arose for a transition into the realm of deep learning to address this specific problem effectively.

The Solution

• Deep Learning Approach

• Tensorflow Implementation

• Productization of the Solution

Through extensive research and development, we engineered a cutting-edge Deep Learning algorithm leveraging FFM (field-aware factorization machines), enabling a more nuanced representation of ad content. Upon rigorous testing, our solution surpassed the baseline models previously employed, showcasing significantly superior performance across various metrics. The seamless deployment and productization of this advanced solution were seamlessly executed within the client’s infrastructure, utilizing Kubernetes clusters to ensure scalability and efficiency. This innovative approach not only revolutionized ad content analysis but also underscored our commitment to delivering state-of-the-art solutions tailored to our clients’ needs.