AI Server Performance Applications Engineer – Benchmarking

Akash Systems; Emeryville (CA); On-site; USD 140,000 - 210,000; Full time

Benefits offered by this job

Health insurance

Dental insurance

Vision insurance

401(k) with company match

Annual performance bonus

Job summary

Akash Systems, Inc. seeks an Applications Engineer, AI Server Software Performance to bridge engineering, customers, and partners.

You will lead training and inference benchmarking, support POC engagements, and contribute to software development initiatives. You will drive performance evaluation across training/inference workloads, own end-to-end deployment on CUDA/ROCm, and help translate hardware capabilities into optimized software solutions for AI workloads.

Qualifications

• Bachelor's degree in Computer Engineering or Computer Science; Master’s degree preferred.
• Minimum of 5 years of relevant industry experience.
• Familiarity with CUDA and RocM software platforms.
• Strong written and verbal communication skills.
• Experience analyzing, interpreting, and presenting technical data.

Responsibilities

• Evaluate AI server performance across training and inference benchmarks.
• Build and maintain a catalog of performance and benchmark data.
• Lead POC demonstrations and customer testing.
• Own end-to-end LLM deployment across CUDA and ROCm platforms.
• Drive internal/external benchmarking initiatives (e.g., MLPerf).
• Bridge OEM partners and chip manufacturers to optimize software solutions.
• Identify and resolve performance bottlenecks across server stack.

Skills

CUDA/RocM familiarity

Communication skills

Project management

Analytical thinking

Education

Bachelor's degree in Computer Engineering or Computer Science

Master’s degree preferred

Tools

CUDA toolkit

ROCm

MLPerf benchmarking

Job description

Akash Systems, Inc. seeks an Applications Engineer, AI Server Software Performance to bridge engineering, customers, and partners.

You will lead training and inference benchmarking, support POC engagements, and contribute to software development initiatives. You will drive performance evaluation across training/inference workloads, own end-to-end deployment on CUDA/ROCm, and help translate hardware capabilities into optimized software solutions for AI workloads.