Machine Learning Engineer, Drive

DoorDash USA•Sunnyvale, CA

4d•$137,100 - $299,300•Remote

About The Position

DoorDash Drive powers deliveries placed through merchants' own channels—including their websites, mobile apps, and phone orders—using DoorDash's logistics network. The Drive Machine Learning team builds the prediction and intelligence systems that power this business, including delivery and pickup time estimation, merchant prep-time prediction, order release optimization, logistics decision-making, and AI-powered delivery quality signals. Drive presents a unique machine learning challenge. Every merchant has different operational workflows, preparation patterns, and customer expectations, requiring models that generalize across millions of deliveries while adapting to highly diverse merchant behavior. Our team has significant opportunities to improve prediction accuracy, optimize logistics decisions, and build AI-native experiences that directly improve merchant, consumer, and dasher outcomes. As a Machine Learning Engineer on the Drive team, you'll own machine learning systems end-to-end—from feature engineering and model development to experimentation, deployment, monitoring, and continuous iteration. Your work will span several high-impact problem areas: * Build next-generation machine learning models for delivery ETA, pickup ETA, merchant prep-time estimation, and order release prediction that improve reliability for merchants and consumers. * Develop deep learning models that leverage large-scale spatiotemporal, marketplace, and behavioral signals to improve prediction accuracy. * Apply reinforcement learning and optimization techniques to improve logistics decision-making, assignment strategies, and marketplace efficiency. * Build AI-native product experiences using large language models (LLMs) and vision-language models (VLMs). For example, transform pickup photos, item verification flows, receipts, and drop-off images into structured quality signals that help verify orders, prevent delivery defects, and improve issue resolution. * Design and run rigorous online experiments, production monitoring, and model iteration to continuously improve performance. * Partner closely with software engineers, product managers, data scientists, and platform teams to bring new machine learning capabilities into production at scale. You'll have the opportunity to work across traditional machine learning, deep learning, reinforcement learning, optimization, and multimodal AI while solving some of the most challenging logistics problems at DoorDash.

Requirements

5+ years of industry experience building and shipping production machine learning systems with measurable business impact (Bachelor's, Master's, or PhD).
Strong experience developing production machine learning models using modern deep learning frameworks such as PyTorch and distributed data processing technologies such as Spark and Airflow.
Experience building, deploying, monitoring, and maintaining production ML systems end-to-end.
Strong software engineering skills in Python and experience with modern ML infrastructure and tooling.
Deep expertise in at least one of the following areas: Deep Learning, Reinforcement Learning, Optimization / Operations Research, Large Language Models (LLMs) or Vision-Language Models (VLMs)
Experience applying machine learning to estimation, ranking, prediction, optimization, or decision-making problems at production scale.
Proficiency using AI-assisted development tools (e.g. Claude Code, Codex, Cursor) throughout the software development lifecycle.
You are located or are planning to relocate to San Francisco, CA, Sunnyvale, CA, or Seattle, WA.

Nice To Haves

Hands-on experience with LLMs or VLMs is a strong plus.
Experience in logistics, marketplaces, or delivery platforms is helpful but not required.

Responsibilities

Build next-generation machine learning models for delivery ETA, pickup ETA, merchant prep-time estimation, and order release prediction that improve reliability for merchants and consumers.
Develop deep learning models that leverage large-scale spatiotemporal, marketplace, and behavioral signals to improve prediction accuracy.
Apply reinforcement learning and optimization techniques to improve logistics decision-making, assignment strategies, and marketplace efficiency.
Build AI-native product experiences using large language models (LLMs) and vision-language models (VLMs). For example, transform pickup photos, item verification flows, receipts, and drop-off images into structured quality signals that help verify orders, prevent delivery defects, and improve issue resolution.
Design and run rigorous online experiments, production monitoring, and model iteration to continuously improve performance.
Partner closely with software engineers, product managers, data scientists, and platform teams to bring new machine learning capabilities into production at scale.