AI Hardware Performance Architect

Sygaldry Technologies, Inc.; San Francisco (CA); On-site; USD 140,000 - 190,000; Full time

Benefits offered by this job

Visa sponsorship

Competitive equity & salary

Health coverage for you and family

Team offsites & social events

Unlimited PTO

Job summary

Sygaldry Technologies, Inc. is seeking researchers to advance quantum-accelerated AI server concepts through high-fidelity performance models and simulators. You will work across compute, memory, networking, and data movement to understand bottlenecks in distributed ML workloads.

The role emphasizes model parallelism, data parallelism, and exploration of hardware/software tradeoffs to accelerate AI training and inference at scale.

Qualifications

• PhD or Masters in Computer Engineering with strong research track record.
• Experience with ML systems and large-scale training/inference.
• Knowledge of distributed computing, performance modeling, and systems benchmarking.

Responsibilities

• Develop performance models and simulators for large-scale ML systems.
• Model training/inference performance across compute, memory, networking, and data movement.
• Analyze workload parallelism and scheduling to identify bottlenecks.
• Explore architectures and tradeoffs to improve performance and scalability.
• Build simulation frameworks for rapid design-space exploration.
• Collaborate with ML researchers and system architects to guide AI system design.

Skills

PhD in CS/CE

Distributed systems

Performance modeling

ML systems

Simulation frameworks

GPU clusters

Data/model parallelism

Benchmarking tools

Education

PhD in Computer Engineering

Masters in Computer Engineering

Tools

Simulation tools

Benchmarking tools

Job description

Sygaldry Technologies, Inc. is seeking researchers to advance quantum-accelerated AI server concepts through high-fidelity performance models and simulators. You will work across compute, memory, networking, and data movement to understand bottlenecks in distributed ML workloads.

The role emphasizes model parallelism, data parallelism, and exploration of hardware/software tradeoffs to accelerate AI training and inference at scale.