About the role#
The Data-AML-Engine Orchestration team builds large-scale machine learning infrastructure that powers online model serving across ByteDance products, including TikTok. You will develop orchestration, scheduling, and resource management systems that connect heterogeneous compute infrastructure with production machine learning workloads. This role directly impacts GPU utilization, serving latency, infrastructure reliability, and developer productivity.
What you'll need#
- Currently pursuing a Bachelor's or Master's degree in Computer Science, Software Engineering, Artificial Intelligence, or a related technical field.
- Proficiency in Go, C++, or Python, with a solid foundation in data structures, algorithms, and software engineering principles.
- Familiarity with Linux and a foundational understanding of operating systems, computer networks, concurrent programming, and distributed systems.
- Strong hands-on abilities and a willingness to investigate systems through source code, metrics, logs, and profiling.
- A systematic approach to problem solving, including the ability to define measurements, test hypotheses, and validate improvements.
- Demonstrated ownership through coursework, research, internships, or open-source projects.
- Experience with Kubernetes, container runtimes, resource scheduling, or multi-tenant systems is a plus.
- Familiarity with model serving systems like vLLM, SGLang, Triton, KServe, or Ray Serve is beneficial.
- Experience with online services, traffic management, or GPU/NPU programming is an advantage.
What you'll do#
- Design and build foundational orchestration for machine learning platforms, including Kubernetes Operators, container runtimes, and lifecycle management for jobs and services.
- Build multi-tenant resource and quota systems that support priorities, preemption, fair sharing, elasticity, and cross-cluster scheduling.
- Improve GPU utilization and cost efficiency through resource pooling and FinOps.
- Create lifecycle orchestration for online model serving, such as model distribution, deployment, upgrades, rollback, autoscaling, and disaster recovery.
- Build serving orchestration and traffic management for disaggregated clusters, including topology-aware scheduling, KV Cache affinity, request routing, and QoS management.
Location & details#
- Location: San Jose, California, United States.
- Employment type: Full-time intern.
- Compensation: $45 per hour.
- Modality: On-site.
- Hiring process: Applications are reviewed on a rolling basis.


