Senior Data Scientist, Ads Modeling
The front page of the internet, Reddit brings over 430 million people together each month through their common interests, inviting them to share, vote, comment, and create across thousands of communities. Come for the cats, stay for the empathy.
We are looking for a Modeling Data Scientist to work within the Ads Data team. You will work closely with engineers and product owners from our Ads team to optimize ads delivery and auction systems. In addition to strong modeling skills, this person has a solid business acumen and understands what is important to advertisers.
- Build optimization algorithms that improve ad yields and efficiencies. Some of the models we developed over the last year are CPC (cost per click), CPI (cost per install), Generalized pCVR, Ad level optimization and User lookalikes. You will not only work on improving existing optimization models, but also develop new models from scratch for a variety of upcoming product launches.
- Build and improve Machine Learning algorithms that match ads to the most relevant users. Some of the algorithms/techniques used are LR, GBDT, RF, Hyper parameter tuning, Thomson Sampling, Monte Carlo simulations, semantic embedding models etc.
- Build a platform for rapid model iteration and feature engineering at scale. Some of our optimization models are iterated on and deployed in production every 2 weeks.
- Be involved in all phases of modeling such as ideation, offline modeling, online implementation, experimentation, deploy and post-launch monitoring/measurements.
- This role will have a lot of overlap with other Machine Learning Engineer roles, but will differ in a couple of areas. First is that you will work mainly on offline modeling and rely on engineers to productionalize your models. Secondly, you will have a keen interest in the collection and quality of underlying data, along with working on ETLs and data aggregations.
- Serve as a thought-partner for Product Managers, Engineering Managers and leadership in influencing the Monetization roadmap and strategy for Reddit by identifying opportunities through deep-dive analyses and/or modeling.
- Deeply understand ad auctions and how bid densities affect CPMs, CPCs and revenue
- Develop familiarity with different types of ads such as managed/reserved, auction, takeover and self-serve and understand specific interplays amongst those ads.
- Work closely with our Sales and Marketing partners so the ads are set up in a way that amplifies the benefits of your optimization models.
What We Can Expect From You:
- Bachelor’s degree or above in a quantitative major (e.g., mathematics, statistics, economics, finance, computer science).
- Proficiency in Machine Learning
- Proficiency with statistical analysis and programming languages (Python, SQL)
- Understanding of experimentation and causal inference analyses
- Master’s or PhD degree preferred but not required.
- 2+ years of experience (for Bachelor’s); 1+ years of experience (for Master’s/Phd) in quantitative/modeling roles, preferably for a consumer-facing service/app
- Experience in online advertising is preferred but not required