MLOps Engineer

## Responsibilities - Maintain and improve our training and evaluation infrastructure — data pipelines, annotation tooling, experiment tracking, and model versioning. - Drive rigorous model evaluation — precision, recall, calibration, and per-pathology sensitivity/specificity trade-offs on real clinical data. - Optimize model inference for production: latency, throughput, quantization, and hardware-efficient deployment. - Contribute to research direction: what to build next, which problems are tractable, and how to measure success. ## Requirements - MSc in a relevant field (computer vision, deep learning, or similar); PhD welcome. - 4+ years of machine learning experience focused on computer vision, with at least 2 years shipping models in production. - Deep proficiency with PyTorch; experience with object detection frameworks (YOLO, DETR, or similar) and inference optimization (ONNX, TensorRT, quantization) a plus. - Track record of designing or significantly modifying model architectures for specific problems — not just fine-tuning off-the-shelf models. - Strong understanding of evaluation methodology — not just benchmark metrics, but sensitivity/specificity trade-offs in high-stakes classification. ## Nice to Have - Familiarity with medical imaging (DICOM, radiograph modalities) is an advantage. ## Benefits - Work at the core of what makes Dentail AI clinically credible — the detection models that identify pathologies and other findings in dental radiographs with accuracy that meets or exceeds specialist performance. - You'll own the full MLOps lifecycle from research to production deployment and continuous evaluation. - Dentail's AI engines are an ensemble of several ML models interacting through interconnected logical layers, so this role rewards both depth in modeling and systems thinking.