About the job
As a Machine Learning Engineer focused on distributed vLLM infrastructure in the llm-d project, you will be at the forefront of innovation, collaborating with our team to tackle the most pressing challenges in scalable inference systems and Kubernetes-native deployments. Your work with machine learning, distributed systems, high performance computing, and cloud infrastructure will directly impact the development of our cutting-edge software platform, helping to shape the future of AI deployment and utilization.
Responsibilities
Contribute to the design, development, and testing of new features and solutions for Red Hat AI Inference
Innovate in the inference domain by participating in upstream communities
Develop and maintain distributed inference infrastructure leveraging Kubernetes APIs, operators, and the Gateway Inference Extension API for scalable LLM deployments.
Develop and maintain system components in Go and/or Rust to integrate with the vLLM project and manage distributed inference workloads.
Develop and maintain KV cache-aware routing and scoring algorithms to optimize memory utilization and request distribution in large-scale inference deployments.
Enhance the resource utilization, fault tolerance, and stability of the inference stack.
Develop and test various inference optimization algorithms.
Actively participate in technical design discussions
Contribute to a culture of continuous improvement by sharing recommendations and technical knowledge with team members
Collaborate with other engineering and cross-functional teams to deliver on engineering deliverables
Communicate effectively to team members to ensure proper visibility of development efforts
Be taught, coached, and mentored by senior members of the team
Provide timely and constructive code reviews
Qualifications
Minimum
Strong proficiency in Python and/or GoLang or similar language
Experience with cloud-native Kubernetes service mesh technologies/stacks such as Istio, Cilium, Envoy (WASM filters), and CNI.
Working understanding of Layer 7 networking, HTTP/2, gRPC, and the fundamentals of API gateways and reverse proxies.
Knowledge of serving runtime technologies for hosting LLMs, such as vLLM, SGLang, TensorRT-LLM, etc.
Excellent written and verbal communication skills, capable of interacting effectively with both technical and non-technical team members.
Ability work independently in a dynamic, fast-paced environment
Preferred
Proficiency in C, C++, or Rust
Experience with the Kubernetes ecosystem, including core concepts, custom APIs, operators, and the Gateway API inference extension for GenAI workloads.
Working knowledge of high-performance networking protocols and technologies including UCX, RoCE, InfiniBand, and RDMA is a plus.
Experience with GPU performance benchmarking and profiling tools like NVIDIA Nsight or distributed tracing libraries/techniques like OpenTelemetry.
Experience in writing high performance code for GPUs and deep knowledge of GPU hardware
Strong understanding of computer architecture, parallel processing, and distributed computing concepts
Bachelor's degree in computer science or related field is an advantage, though we prioritize hands-on experience
Active engagement in the ML research community (publications, conference participation, or open source contributions) is a significant advantage