5 of 5 Permanent Amazon SageMaker Jobs in Glasgow

including model hosting, request routing, continuous batching, and KV-cache optimization Deploy, host, and lifecycle-manage open-source and proprietary LLMs on Amazon EKS and Amazon SageMaker, as well as on-prem and local GPU clusters, using reproducible infrastructure as code and continuous delivery pipelines Implement observability … experience operating LLM inference servers such as vLLM and llm-d (or directly equivalent serving stacks) Hands-on experience hosting and serving LLMs on Amazon EKS and/or Amazon SageMaker, and on local GPU infrastructure Knowledge of LLM reliability and risk considerations, including latency/throughput ...

integrating model serving platforms such as KServe, Ray Serve, or NVIDIA Triton Inference Server alongside other large language model serving stacks Familiarity with Amazon SageMaker JumpStart, SageMaker Endpoints, and Amazon Bedrock for managed model hosting Experience with online large language model quality monitoring, including hallucination detection ...

operating or integrating model serving platforms such as KServe, Ray Serve, or NVIDIA Triton Inference Server alongside other large language model serving stacksFamiliarity with Amazon SageMaker JumpStart, SageMaker Endpoints, and Amazon Bedrock for managed model hostingExperience with online large language model quality monitoring, including hallucination detection ...

building agentic AI systems (multi-step workflows, tool routing, planning, memory patterns, supervision/fallback strategies). Experience with AWS Bedrock and/or SageMaker (or equivalent managed ML/GenAI platforms) and deployment patterns for scalable inference. Experience with evaluation frameworks and approaches (golden datasets, LLM-as-judge ...

processing, feature engineering, model selection, hyper-parameter tuning, and evaluation, including choosing appropriate metrics for imbalanced and unlabeled problems. Experience with AWS Bedrock, SageMaker (or equivalent managed ML/GenAI platforms), ECS and deployment patterns for scalable inference. Experience with developer productivity tooling such as GitHub Copilot and Claude ...