AI Systems Co-Design Research Scientist

Meta; New York (NY); On-site; USD 170,000 - 230,000; Full time

Job summary

Meta seeks a PhD-level researcher to advance hardware-aware AI model enablement and performance modeling for AI accelerators. The role emphasizes end-to-end architecture, on-device mapping, and working with CNNs, Transformers, LLMs, and diffusion models.

The candidate will analyze system-level performance, collaborate with cross-functional teams, and publish results at top venues while contributing to production-grade AI workloads and optimization strategies.

Qualifications

• Proven research experience in hardware-aware model enablement or AI accelerator performance modeling.
• Hands-on end-to-end AI hardware architecture and on-device mapping algorithm development.
• Background with CNNs, Transformers, LLMs, or diffusion models.
• Experience in system-level performance analysis, profiling, benchmarking AI workloads.
• Proficiency in Python and at least one major AI framework.
• Track record of publishing research and communicating results cross-functionally.
• Experience deploying AI agents or pretrained techniques for efficiency.
• Experience or knowledge of large-scale training/inference.
• Familiarity with low-level programming for specialized hardware or HDL.
• Experience or knowledge of distributed ML systems.
• Knowledge of Generative AI models or Ranking/Recommendation models.

Skills

Python

AI frameworks

Performance modeling

System-level analysis

Distributed ML

Research publication

Communication

CUDA

Triton

LLMs/Transformers

Education

PhD in Computer Science or Electrical Engineering

Master's +3 years industry

Tools

CUDA

HDL

Triton

DeepLearning frameworks

Job description

Meta seeks a PhD-level researcher to advance hardware-aware AI model enablement and performance modeling for AI accelerators. The role emphasizes end-to-end architecture, on-device mapping, and working with CNNs, Transformers, LLMs, and diffusion models.

The candidate will analyze system-level performance, collaborate with cross-functional teams, and publish results at top venues while contributing to production-grade AI workloads and optimization strategies.