MLOps Engineer

Svitla Systems Inc. is looking for an MLOps Engineer for a full-time position (40 hours per week) in Romania. Our client develops intelligent devices and systems for real-time supply chain, robotics, and computer vision operations.

Requirements Programming & Development:

• Advanced proficiency in Python development
• Working experience with PyTorch and/or TensorFlow
• Ability to create maintainable, well-tested code with proper error handling
• Experience creating RESTful services and model serving endpoints

MLOps Infrastructure

• Design and implementation of automated ML workflows (training, evaluation, deployment)
• Experience with model versioning, storage, and metadata management
• Implementing dataset versioning and change management
• Knowledge of model optimization and format conversion (ONNX, TFLite, TensorRT, SNPE)

DevOps & Infrastructure

• Docker image building and optimization
• Experience with Kubernetes and/or RunAI
• GitHub Actions workflow design and implementation
• GCP experience, particularly with Vertex AI, Cloud Batch, or similar services
• GPU allocation and optimization for training workloads

Testing & Quality Assurance

• Unit, integration, and E2E test implementation for ML systems
• Model metrics calculation and performance benchmarking
• Experience with testing ML models on edge devices

Additional Technologies

• Experience with Hydra or similar configuration frameworks
• Familiarity with W&B, MLflow, or similar tracking tools
• Creating and maintaining Python packages and dependencies
• Understanding of CV models, image processing, and related ML tasks

Responsibilities

• Design and implement an end-to-end MLOps architecture for computer vision models
• Create automated pipelines for model training, quantization, conversion, and evaluation
• Develop a model registry with comprehensive versioning and metadata capabilities
• Build systems for automated model testing across multiple target architectures
• Implement data versioning and dataset management solutions
• Maintain CI/CD pipelines for ML model lifecycle management
• Optimize Docker environments for development and deployment
• Develop Python libraries and APIs for internal model consumption