Senior Machine Learning Engineer (Ads Response Prediction)

Role Overview

Instacart is seeking a Senior Machine Learning Engineer II for the Ads Response Prediction team. This research-focused role will help lead the design and development of core ML models powering Instacart’s ads ecosystem.

Key Responsibilities

• Design and develop core ML models for ads response prediction, including theoretical problem formulation and ensuring strong model quality.
• Improve pCTR modeling and model calibration across different surfaces and domains.
• Research and implement bias-mitigation methods for selection bias and position bias (including propensity-based correction approaches).
• Advance multi-task learning and sequence modeling capabilities through collaboration with the ML community.
• Shape the next-generation foundation model approach for ads ranking.
• Contribute to retrieval systems (including generative retrieval / next-gen ranking architectures).

Requirements

• PhD or Master in Machine Learning, Statistics, Computer Science, Information Retrieval, or a closely related quantitative field.
• Strong foundation in causal inference, counterfactual reasoning, and training data bias mitigation, including reasoning about selection bias and position bias.
• Proficiency in Python and deep learning frameworks: PyTorch, TensorFlow, or JAX.
• Fluency with data tools: SQL, Spark, Pandas.
• Strong communication skills; able to explain modeling decisions to cross-functional stakeholders (e.g., product managers and data scientists).
• Deep understanding of CTR / conversion prediction; familiarity with Deep & Wide, DeepFM, DCN, and multi-task learning formulations.
• 6+ years combined academic/industry experience applying ML to ranking/recommendation/prediction at scale (including PhD research).
• Proven ability to turn ambiguous problems into well-scoped ML research directions and deliver via rigorous experimentation.

Nice-to-haves

• Experience with ads ranking / auction-based systems (e.g., pCTR, bid optimization, ROAS feedback loops, marketplace dynamics).
• Hands-on experience with autoregressive sequence models, generative retrieval, or transformer-based ranking.
• Familiarity with learned representations such as Semantic IDs and product embeddings (including feature cardinality reduction and cold-start mitigation).
• Publication record in venues such as KDD, WWW, RecSys, NeurIPS, ICML, or SIGIR.
• Experience with transfer learning / domain adaptation (e.g., LoRA, adapter-based fine-tuning).
• Experience mentoring junior engineers and shaping modeling team technical direction.
• Familiarity with LLM-driven recommendation (prompt-based personalization, AI-assisted model development / AutoML).