Applications Engineer, AI Server Software Performance

Company Description

Akash Systems, Inc. is a venture-backed, late-stage, Bay Area company that makes and sells Diamond Cooled GPU-based servers to AI companies, Cloud Service Providers, and data centers worldwide.  The deep tech company pioneered Diamond Cooling® technology wherein the world’s most thermally conductive material, lab-grown diamond, is brought close to the GPU chip – the densest heat source in a modern high compute server. The resulting Akash server exhibits more compute in FLOPS (>50%) and FLOPS per Watt (by 2x) than any other server in the market today. Akash’s Diamond Cooled servers also reduce the energy cost of cooling the entire data center and maintain performance in high ambient temperatures. The company’s lead investors include Khosla Ventures and Founders Fund. Akash Systems has deployed its Diamond Cooling® technology in space, with many Akash satellite radios in orbit. 

Role Description

We are seeking an Applications Engineer, AI Server Software Performance to serve as the technical bridge between our engineering teams, customers, and strategic partners. This individual will lead training and inference benchmarking activities, support customer proof-of-concept (POC) engagements, and contribute to software development initiatives.  

Responsibilities

• Evaluate performance of AI servers across training and inference testing benchmarks
• Build and maintain catalog of performance and benchmark data across benchmark models
• Lead implementation and support customer testing / proof of concept demonstrations of inference and training models on servers
• Own E2E LLM deployment across NVIDIA and AMD platforms, from kernel optimization on CUDA and ROCm to production ready inference stacks.
• Drive internal and external testing and benchmarking initiatives, including 3rd party benchmarks like MLCommons’ MLPerf Inference and MLPerf Training workloads.
• Serve as a technical bridge between OEM partners and chip manufacturers (e.g., NVIDIA, AMD, Supermicro, Dell), translating hardware capabilities into optimized software solutions for training and inference workloads.
• Identify and resolve performance bottlenecks across the full server stack, from driver and firmware compatibility to model quantization (FP8/FP4) and throughput tuning.

Required Qualifications

• Bachelor's degree in Computer Engineering or Computer Science; Master’s Degree preferred. 
• Minimum of 5 years of relevant industry experience following completion of degree. 
• Strong Python programming skills. 
• Familiarity with CUDA and RocM software platforms. 
• Strong written and verbal communication skills. 
• Experience analyzing, interpreting, and presenting technical data to both internal and external reporting. 
• Excellent project management skills. 

Compensation & Benefits

We offer a competitive compensation package commensurate with experience, including:

· Base salary: competitive and based on experience, skills, and qualifications

· Annual performance bonus

· Comprehensive health, dental, and vision insurance

· 401(k) with company match

· Flexible PTO policy

Work Authorization

Akash Systems does not sponsor or take over sponsorship for employment visas for this role (e.g., H‑1B, O‑1, TN, F‑1/OPT, etc.). Eligible candidates are U.S. citizens or U.S. lawful permanent residents (Green Card Holders).