Senior ML Engineer - Ads Forecasting & Scenarios

Fdetimes; United States; Remote; USD 217,000 - 303,000; Full time

Benefits offered by this job

Healthcare benefits

401k with employer match

Parental leave

Flexible vacation

Global benefits and development

Job summary

Reddit is seeking a senior forecasting scientist on the Forecasting team to own end-to-end predictive ML systems that estimate campaign outcomes under budget, bid, and audience changes. You will connect supply and demand signals, forecast delivery and response, and drive production deployments with engineers and partner teams.

You will translate advertiser questions into modeling problems, build robust forecasts, evaluate uncertainty, and iterate to improve calibration across advertiser cohorts.

Qualifications

• Experience independently designing, shipping, and improving predictive ML systems in advertising, marketplaces, recommendations, search, forecasting, or similarly complex domains.
• Strong foundations in statistical learning, probability, and optimization, with experience working with large-scale behavioral or transactional data.
• Practical experience in at least one relevant area, such as forecasting, response prediction, probabilistic modeling, multitask learning, or transfer learning.
• Sound evaluation judgment, including temporal validation, segment-level analysis, and an understanding of calibration, uncertainty, and the limits of predictions under changed conditions.
• Experience diagnosing and addressing modeling challenges such as sparse observations, cold start, seasonality, distribution shift, selection bias, or delayed labels.
• Strong programming and software engineering skills, including experience building and operating production ML pipelines and services.
• Demonstrated ability to turn ambiguous product questions into concrete technical plans, lead projects through rollout, and collaborate effectively across functions.

Responsibilities

• Own forecasting and scenario modeling projects. Translate advertiser questions into modeling problems involving delivery, reach, performance, and investment tradeoffs.
• Connect inventory, demand, and response. Use signals about available advertising opportunities, audience activity, marketplace competition, and predicted response to improve estimates of achievable campaign outcomes.
• Model outcomes under alternative choices. Build and improve models that estimate the effects of changes in budget, bid, audience, or optimization objective, accounting for relevant auction dynamics and campaign constraints.
• Help advertisers identify opportunity. Incorporate predictions from partner teams, including creative-performance estimates, to evaluate potential improvements in delivery, efficiency, or advertiser value.
• Improve model generalization. Develop approaches for sparse segments, cold start, seasonality, delayed feedback, and changing marketplace conditions. Choose methods that balance predictive quality, complexity, and production requirements.
• Build trustworthy evaluations. Measure forecast accuracy, calibration, uncertainty, and performance across advertiser cohorts. Document modeling assumptions and limitations, and work with partners to determine which scenario comparisons the evidence supports.
• Close the learning loop. Compare predictions with mature outcomes, investigate systematic errors, and use experiments and observed results to improve forecasts and validate scenario estimates.
• Deliver reliable production systems. Own training, inference, monitoring, and iteration for your models. Collaborate with Product, Data Science, Ads Delivery, Audience, and Creative teams; contribute to design and code reviews; and mentor engineers through project work.

Skills

Forecasting

Statistical learning

Probability

Optimization

Production ML pipelines

Tools

Python

SQL

TensorFlow

PyTorch

Airflow

Job description

Reddit is seeking a senior forecasting scientist on the Forecasting team to own end-to-end predictive ML systems that estimate campaign outcomes under budget, bid, and audience changes. You will connect supply and demand signals, forecast delivery and response, and drive production deployments with engineers and partner teams.

You will translate advertiser questions into modeling problems, build robust forecasts, evaluate uncertainty, and iterate to improve calibration across advertiser cohorts.