Source Job

$140,000–$150,000/yr
Global

  • Design and execute performance benchmarks for AI training and inference workloads.
  • Profile and characterize GPU workloads to identify bottlenecks and optimization opportunities.
  • Systematically tune workload parameters to maximize throughput and establish performance baselines across GPU platforms.

Python

10 jobs similar to GPU Performance and Benchmarking Engineer

Jobs ranked by similarity.

$125,000–$135,000/yr
Global

  • Validate and troubleshoot InfiniBand and RoCE fabrics during GPU cluster bring-up and expansion.
  • Tune fabric performance parameters for distributed AI workloads such as NCCL and MPI.
  • Collaborate with GPU and networking teams to diagnose and resolve fabric-level issues and optimize performance.

Vultr makes high-performance cloud infrastructure easy to use and affordable for enterprises and AI innovators worldwide. With 33 global data centers and hundreds of thousands of customers, it is the largest privately-held cloud infrastructure company, offering a culture of innovation and growth.

  • Optimize production LLM serving with vLLM and SGLang to maximize throughput and minimize latency through batching and quantization.
  • Profile training runs to find bottlenecks and resolve them with attention implementations like FlashAttention on H200 and GB200 hardware.
  • Deploy and operate multiple models on shared GPU clusters with autoscaling, bin-packing, and efficient handling of mixed workloads.

Egen is a fast-growing technology company with a data-first mindset, partnering with clients on Google Cloud and Salesforce to drive action through data and insights. We are a team of dedicated engineers who thrive on solving tough problems and continually innovate to achieve fast, effective results.

$125,000–$200,000/yr
Global Unlimited PTO

  • Implement and maintain AI benchmarks using evaluation infrastructure like the Inspect library.
  • Contribute to the design and development of new benchmarks for frontier AI models.
  • Collaborate with researchers and engineers to ensure accurate and insightful evaluation data.

Epoch AI is a research institute investigating trends in machine learning and the economic consequences of AI. With a small, mission-driven team, we aim to provide rigorous, independent insights into AI development.

United States

  • Lead engineering deployment, scaling, and operations of AI compute clusters with GPU fleets and bare metal environments.
  • Drive reliability, monitoring, automation, and incident response for AI infrastructure.
  • Collaborate with AI/ML, networking, and product teams to align infrastructure with business needs.

Our partner is a fast-growing cloud environment focused on building large-scale AI infrastructure. They seek a senior leader to manage engineering operations for advanced AI compute clusters.

Europe

  • Own performance optimization and reliability of large-scale GPU clusters and InfiniBand networking for HPC workloads.
  • Diagnose and resolve complex system-level issues across GPU, network, and compute layers, integrating new hardware components.
  • Develop automation for monitoring, fault detection, and proactive remediation in distributed compute environments.

Our partner is building a next-generation AI cloud infrastructure environment, focusing on large-scale high-performance computing systems. They foster a highly technical engineering culture with experts across systems, networking, and virtualization, offering career development and continuous learning opportunities.

$225,000–$325,000/yr
Global Unlimited PTO

  • Lead core infrastructure and SRE teams to ensure the highest reliability and performance for massive GPU computing demands.
  • Oversee HPC networking and distributed storage engine innovation to support massive multi-node AI workloads.
  • Build and scale a high-output engineering org while partnering cross-functionally with product and GTM leadership.

Runpod is the AI Developer Cloud, providing a platform for over one million developers to experiment, train, fine-tune, deploy, and scale AI. As a small, remote-first team that closed a $100M Series A, we move fast and take ownership seriously.

US 20w maternity 12w paternity

  • Own end-to-end technical execution for strategic customer and partner engagements, including discovery, infrastructure design, implementation, and production deployment.
  • Design and build cloud infrastructure supporting advanced AI workloads, including simulation, training, evaluation, inference, and large-scale batch processing.
  • Improve platform reliability, security, performance, and cost efficiency by debugging issues across application, network, storage, compute, and orchestration layers.

The partner company is building the infrastructure foundation for next-generation AI applications and physical AI workloads. The engineering team is pioneering and values ownership, technical excellence, and solving challenging engineering problems at scale.

Canada

  • Lead the design and operation of GPU infrastructure for AI workloads.
  • Manage Kubernetes-based environments and optimize for AI training and inference.
  • Define operational standards, implement monitoring, and collaborate with AI engineering teams.

ELEKS is a software engineering company that partners with enterprises to accelerate digital transformation. They have a global team of over 2,000 professionals and foster a culture of innovation and collaboration.

$100,000–$180,000/yr
US Unlimited PTO

  • Participate in sales meetings, provide architectural recommendations, and build proof-of-concept solutions for onboarding high-spending customers.
  • Troubleshoot and resolve complex technical issues using code analysis, scripting, and log analysis.
  • Create and maintain technical documentation and deliver training sessions, webinars, and demos.

Runpod is the AI Developer Cloud, providing a platform for over one million developers to experiment, train, fine-tune, deploy, and scale AI. We are a small, remote-first team that takes ownership seriously, moves fast, and ships work relied on by more than a million developers daily.

EMEA

  • Build and operate production-grade model serving infrastructure using vLLM, TGI, or Triton frameworks.
  • Design and implement auto-scaling, multi-model architectures, and intelligent request routing for ML inference.
  • Optimize GPU utilization, memory efficiency, and observability to ensure low-latency, cost-effective systems.

They are a distributed cloud infrastructure startup building AI-native cloud services with GPU-powered compute. The company is well-funded, fast-scaling, and operates in a remote-first environment with a focus on sustainability and decentralization.