#MLOpsvacancy #MLOpsjob #MLOpsengineer #vacancy

Senior MLOps ❇️ | Company Top Selection

🔥 We are looking for** an MLOps engineer for project-based employment **

Grade: Senior Rate: from 264K to 290K Citizenship/Location: RF **Workload: **full-time Term: long-term Employment type: only individual entrepreneur 📌 (not civil service, not TK)

✔️Key Requirements Practical experience in building MLOps processes for projects using neural networks. Experience with PyTorch. Experience in automating model training using ClearML, ClearML SDK, ClearML Agent. Experience in converting models to ONNX. Practical experience in inference optimization using TensorRT. Understanding and experience in applying model quantization methods (FP16, INT8, FP8). Experience in deploying and supporting inference services using Triton Inference Server, ClearML Serving, or similar solutions (e.g., BentoML). Experience with vLLM for serving LLM models will be an advantage. Experience with Kubernetes and Helm. Practical experience in managing GPU resources and organizing GPU scheduling. Experience with ML task orchestration tools such as Kubeflow, Ray, KServe.

➕Will be a plus Experience in operating ML infrastructure in both cloud environments and On-Prem. Experience in optimizing the performance of GPUs and inference of large models. Understanding of modern CI/CD practices for ML (MLOps).

📌Main Tasks Development and support of MLOps infrastructure for training, deployment, and maintenance of machine learning models. Automation of neural network training processes, experiment management, and training pipelines. Preparation and optimization of models for industrial inference. Deployment and maintenance of inference services in cloud and On-Prem infrastructure. Configuration and maintenance of GPU clusters, ensuring efficient use of computing resources. Implementation and support of CI/CD processes for ML models. Interaction with the Data Science team for the implementation of models into industrial operation. Monitoring of model and infrastructure performance, optimization of training and inference time.