Machine Learning Engineer, PhD Intern

Instacart

21h•Remote

About The Position

Instacart is seeking talented Ph.D. students to join its fast-moving Machine Learning (ML) teams. The role involves working on high-impact problems at the intersection of Large Language Model (LLM) research, large-scale ML systems, and real-world e-commerce applications. The intern will contribute to transforming the grocery industry by leveraging ML and Internet-scale data to enhance customer experience, improve efficiency, and reduce costs. Specific areas of focus include query understanding, search relevance and ranking, generative recommendations, LLM evaluation and AIQA systems, low-latency and scalable LLM systems, knowledge graphs, and sequence modeling for user behavior prediction.

Requirements

Ph.D. student in computer science, mathematics, statistics, economics, or related areas.
Strong programming (Python, Golang) and algorithmic skills.
Solid foundations in machine learning, algorithms, or optimization.
Curious, self-motivated, and comfortable working on open-ended problems.

Nice To Haves

Ph.D. student at a top tier university in the United States.
Hands-on experience with generative or traditional modeling frameworks (PyTorch, Tensorflow, vLLM).
Prior industry or research internship in machine learning or AI.
Interest and experience in translating research ideas into scalable production systems.

Responsibilities

Work on high-impact problems at the intersection of LLM research, large-scale ML systems, and real-world e-commerce applications.
Contribute to areas such as query understanding, search relevance and ranking, generative recommendations, LLM evaluation and AIQA systems, low-latency and scalable LLM systems, knowledge graphs, and sequence modeling for user behavior prediction.
Leverage cutting-edge AI and LLM-based techniques to understand user intent, refine queries, and support downstream retrieval and ranking.
Improve search relevance by incorporating signals from user behavior, catalog knowledge, and generative models.
Develop scalable feedback and reward modeling approaches for closed-loop learning (RFT).
Build LLM-based evaluation frameworks to improve the quality and reliability of generative and agentic systems.
Research techniques to deploy LLMs in high-traffic, latency-sensitive production environments.
Work on graph data management and knowledge discovery over grocery catalogs.
Build temporal models for user behavior prediction.