MLOps Engineer

Airobotics develops autonomous drone-in-a-box solutions deployed in demanding real-world environments. We're looking for a hands-on MLOps Engineer to own our ML lifecycle - from data pipelines and training infrastructure to edge deployment and production monitoring - ensuring our AI models perform reliably on autonomous drone platforms in the field.

• Design, build, and maintain end-to-end ML pipelines: data ingestion, preprocessing, versioning, training, evaluation, and deployment
• Manage training infrastructure on cloud platforms (Azure, AWS, RunPod) and track experiments using ClearML
• Implement CI/CD pipelines for ML models with automated testing and safe rollout
• Own the model optimization pipeline for edge deployment on NVIDIA Jetson (TensorRT, ONNX, quantization)
• Build production monitoring for model performance, data drift, and degradation detection in field deployments
• Maintain annotation workflows and data quality standards for computer vision datasets
• Work closely with computer vision researchers to move models from research to production

• Strong Python skills and experience with PyTorch or TensorFlow
• Proven experience building ML pipelines (ClearML, MLflow, DVC, Airflow, or similar)
• Hands-on experience deploying and optimizing models on NVIDIA Jetson or similar edge hardware
• Familiarity with TensorRT, ONNX, and model quantization (INT8/FP16)
• Experience with DeepStream and Tritron.
• Experience with Docker, Kubernetes, and cloud platforms (Azure preferred)
• Experience with databases relevant to ML systems: SQL, NoSQL (MongoDB, Redis), time-series (InfluxDB, TimescaleDB), and vector (Pinecone, pgvector)
• Proficiency with Linux and Bash/shell scripting in production environments
• Experience with data labeling tools (CVAT, Label Studio, or similar)
• Background in computer vision (object detection, tracking, segmentation) and familiarity with relevant evaluation metrics (mAP, precision/recall) - an advantage
• Experience in autonomous systems, robotics, or UAVs - an advantage
• Experience with gstreammer or ffmpeg – advantage