Skip to content
GPU Cluster Resource Scheduling and Optimization Engineer
| Company | Together AI |
|---|
| Location | San Francisco, CA, USA |
|---|
| Salary | $160000 – $230000 |
|---|
| Type | Full-Time |
|---|
| Degrees | |
|---|
| Experience Level | Senior |
|---|
Requirements
- 5+ years of experience in resource scheduling, distributed systems, or large-scale machine learning infrastructure.
- Proficiency in distributed computing frameworks (e.g., Kubernetes, Slurm, Ray).
- Expertise in designing and implementing resource allocation algorithms and scheduling frameworks.
- Hands-on experience with cloud platforms (e.g., AWS, GCP, Azure) and GPU orchestration.
- Proficient in Python, C++, or Go for building high-performance systems.
- Strong understanding of operational research techniques, such as linear programming, graph algorithms, or evolutionary strategies.
- Analytical mindset with a focus on problem-solving and performance tuning.
- Excellent collaboration and communication skills across teams.
Responsibilities
- Develop and implement intelligent scheduling algorithms tailored for distributed AI workloads on multi-cluster and multi-tenant environments.
- Ensure efficient allocation of GPUs, TPUs, and CPUs across diverse workloads, balancing resource utilization and job performance.
- Design optimization techniques for dynamic resource allocation, addressing real-time variations in workload demand.
- Implement load balancing, job preemption, and task placement strategies to maximize throughput and minimize latency.
- Build systems that efficiently scale to thousands of nodes and petabytes of data.
- Optimize training and inference pipelines to reduce runtime and cost while maintaining accuracy and reliability.
- Build tools for real-time monitoring and diagnostics of resource utilization, job scheduling efficiency, and bottlenecks.
- Leverage telemetry data and machine learning models for predictive analytics and proactive optimization.
- Collaborate with researchers, data scientists, and platform engineers to understand workload requirements and align resource management solutions.
- Stay updated with the latest trends in distributed systems, AI model training, and cloud-native technologies.
Preferred Qualifications
- Experience with AI/ML frameworks (e.g., TensorFlow, PyTorch, JAX).
- Familiarity with AI-specific workloads like DDP, sharded training, or reinforcement learning.
- Knowledge of auto-scaling and cost-optimization strategies in cloud environments.
- Contributions to open-source scheduling or orchestration projects.
