Meta/Facebook, AI and Systems Co-design

Position ID:

Meta/Facebook-AI and Systems Co-design-RSMPK [#31789]

Position Title:

Research Scientist

Position Type:

Postdoctoral

Position Location:

Menlo Park, California 94025, United States of America

Subject Area:

AI/Machine Learning

Appl Deadline:

(posted 2026/03/09, listed until 2027/03/08)

Position Description:

tl;dr Meta’s AI and Systems Co-design team is looking for candidates for various roles: postdocs and visiting professors, as well as fresh/experienced PhDs.

Overview

Our org's charter is to do interdisciplinary research across AI hardware, software, and algorithms.

AI Accelerators, ML Algorithms, AI Frameworks, and Model-System Codesign have been focus areas in the past - we are always excited to sponsor new initiatives. Meta connects the world using hyperscale computing - these systems will be your testbeds.

One key difference from traditional research labs is that we own technology transfer - we work closely with infra and product groups to take the most promising research to production. Ultimately, our work drives performance, efficiency, scalability, and model quality across Meta.

Details Our Mandate

Our org owns Meta's hardware tech strategy for AI - finding innovative hardware for GPUs and Meta's custom AI chips, as well as CPU, memory, and storage as well as getting these to work in Meta's hyperscale fleet.

We target a broad range of AI workloads: LLMs, GenAI, recommender systems, and more. We do full-stack optimization - ML Hardware, ML Systems, ML Algorithms, and AI Co-design.

Our Unique Research Philosophy

Our org consists mostly of PhDs, including many-world class researchers. We publish at the best conferences in Architecture (ISCA, HPC, Micro, SC), Systems (OSDI, SOSP, ASPLOS), and ML (Neurips, MLSys, ICML); have won multiple best-paper awards; and contribute extensively to OSS, notably PyTorch. Our work is showcased at aisystemcodesign.github.io

We differ from traditional research labs in key ways:

Direct ownership: Similar to traditional research labs, we influence by building strong partnerships with diverse teams. However, we also take direct ownership of the hardware tech strategy for Meta specifically, and landing innovations inside Meta more broadly. This enables us to lead in many areas while fostering seamless partnerships in others.

Production Systems: Our primary goal is to develop forward-looking innovations in AI, hardware, and software; and co-develop them into production. Our lab is Meta’s multibillion user base and multimillion server fleet. This is in stark contrast to traditional research labs that rely on throw-over-the-wall technology transfer.

Direct Impact: We save Meta tens of billions of dollars in hardware cost, and directly develop innovations that help Meta connect the world. Our work is widely acknowledged within the company and throughout the industry.

Who We Want

Historically, some of the most successful people who have joined us were fresh PhDs with tenure-track offers, but who wanted to take a gap year to gain industry insights on research before starting a faculty job—we highly welcome this type of candidate.

We have also worked successfully with established professors, but they tend to be more hands-on. You will not have an army of graduate students working for you. Most of your collaborators will be peers of equal standing.

For postdocs and visiting professors, contracts can vary from 9 to 24 months, depending on the candidate's availability. Candidates are expected to relocate to Menlo Park, California.

CV and questions should be directed to Adnan Aziz, adnanaziz@meta.com

Application Materials Required:

Submit the following item online at this website to complete your application:

Curriculum Vitae

And anything else requested in the position description.

Further Info:

aisystemcodesign.github.io

1 Hacker Way, Menlo Park, CA 94025