Senior Machine Learning Engineer II, Ads Response Prediction

Instacart is transforming the grocery industry. This role is on the Ads Response Prediction team, which owns all systems, algorithms and ML models to ensure a relevant and engaging Ads experience. It is a research-leaning role focused on theoretical problem formulation, training methodology, and model quality rather than infrastructure or full-stack engineering. The team has strong ML infrastructure and MLOps support, including Delta/DBT-Spark data pipelines, Ray-based distributed training, and automated model deployment.

• Lead research and development of pCTR and conversion prediction models, improving calibration and reducing training data biases (selection bias, position bias, optimizer's curse).
• Design debiasing techniques such as Mixed Negative Sampling (MNS), Inverse Propensity Weighting (IPW), counterfactual risk minimization, and calibration methods (Platt scaling, isotonic regression).
• Contribute to the next-generation Multi-Domain Multi-Task (MDMT) model architecture with Mixture-of-Experts (MoE), Transformer layers, and LoRA adaptors.
• Drive sequence modeling initiatives including the TIGER generative retrieval system and Semantic ID representation learning.
• Collaborate on the path toward Foundation Models using autoregressive user behavior prediction.

• PhD/Master in machine learning, statistics, computer science, information retrieval, or a closely related field.
• 6+ years of combined academic and industry experience applying ML to ranking, recommendation, or prediction problems at scale.
• Deep understanding of CTR/conversion prediction modeling, causal inference, and training data bias mitigation.
• Proficiency in Python and deep learning frameworks (PyTorch, Tensorflow, JAX); fluency in SQL, Spark, Pandas.
• Strong written and verbal communication skills.

Base pay ranges vary by permanent work location: $201,000—$253,500 USD for US-based candidates. Eligible for a new hire equity grant as well as annual refresh grants.

Apply via the job posting link.