Applied ML Scientist: Memory, Graphs & RL for Ads

Meta; Sunnyvale, Menlo Park (CA, CA); On-site; USD 55,000 - 96,000; Full time

Job summary

Meta's ML Innovation for Products team seeks interns to advance state-of-the-art ML for ads systems. You will work across memory-driven representations, large-scale graphs, and multimodal fusion to improve retrieval quality and efficiency.

The role blends research with production-ready tooling, within a collaborative Bay Area/Sunnyvale environment. The team operates like an applied research lab with a production deploy button, and you will contribute to systems powering Facebook, Instagram, and

Qualifications

• Pursuing or has a PhD in ML/AI/CS or related technical field.
• Must have work authorization in country of employment.
• Experience with DL frameworks (PyTorch or TensorFlow).
• Experience with Python, C++, C, or Java.
• Experience researching and building ML/DL systems.
• Intent to return to degree program after internship.
• Proven track record of research results (grants, publications, or patents).
• Software engineering experience via internship, work, or open-source contributions.
• Strong quantitative problem-solving skills.
• Experience with large-scale, high-dimensional data for ML.
• Cross-functional collaboration experience.

Responsibilities

• Develop memory modules for scalable ads recommendation with LFM and EAS workstreams.
• Advance graph learning at billion-node scale with self-serve platform.
• Combine semantic content with graph signals in LLM and multimodal fusion.
• Build an agentic AI layer and IDEA framework for memory, planning, and tools.
• Design RL-based production policies to optimize retrieval flows.

Skills

Research experience

Analytical problem solving

Cross-functional collaboration

Internship experience

Education

PhD degree in ML/AI/CS (or in progress)

Tools

PyTorch

TensorFlow

Python

C++

Java

GitHub

Job description

Meta's ML Innovation for Products team seeks interns to advance state-of-the-art ML for ads systems. You will work across memory-driven representations, large-scale graphs, and multimodal fusion to improve retrieval quality and efficiency.

The role blends research with production-ready tooling, within a collaborative Bay Area/Sunnyvale environment. The team operates like an applied research lab with a production deploy button, and you will contribute to systems powering Facebook, Instagram, and