Data Scientist for Machine Learning Team

Smadex is a performance advertising engine for mobile apps, enabling the world's leading marketers to grow across Mobile and Connected TV (CTV). Powered by AI models trained on real campaign outcomes, Smadex delivers profitable growth at scale. Headquartered in Barcelona, with teams across the Americas, EMEA, and APAC, Smadex is a business unit of Entravision (NYSE: EVC).

Our mission is to continue to improve our ad-tech platform, especially the core of it, our algorithms, to help our clients achieve their programmatic advertising campaign goals. We want to give our employees a job they’ll love, where they will be challenged to improve results through real-life engineering and data analysis and where everyone’s implication has an impact.

Are you ready to be part of the new unicorn? Keep reading!

Ad-tech is one of the most demanding environments for machine learning. At its core, our DSP operates as an ultra-low-latency recommender system, matching the right ad to the right user in milliseconds. You will be operating at massive scale under strict latency constraints, solving unique modeling challenges that aren't even covered in current scientific literature.

As a Data Scientist embedded within our Machine Learning team, your mission is to bridge the gap between theoretical model design and real-world production reality. You will be the analytical anchor dedicated to ensuring the performance, interpretability and stability of our core Deep Neural Networks models within our high-frequency, complex auction ecosystem with millions of ad requests processed every second.

We are looking for a deeply analytical individual with very strong ML expertise—someone who isn't satisfied with "it works," but needs to understand the system at a fundamental level. You will apply scientific rigor to drive measurable improvements in our live bidding performance.

Your Tasks and Responsibilities:

• Model Understanding and Training: Understand our ML models in production at a fundamental level, explaining why they make specific predictions and, behave in the way they behave. To do so, you will autonomously formulate different hypotheses and validate them by performing the appropriate data analyses and by training different models to compare the results.
• Deep-Dive Auction Dynamics Analysis: Conduct in-depth analysis of model behavior in the live auctioned environment to understand how our algorithms and models interact with market dynamics.
• Experimentation: Design and analyze experiments, A/B testing, to validate hypotheses regarding bidding strategies and model improvements.
• System Optimization: Identify performance bottlenecks in the models and propose architectural or feature enhancements to resolve non-optimal performance at the system level.