Machine Learning Engineer

About the role

Jumio is seeking a Senior or Staff level engineer to lead the development and scaling of biometric face recognition systems. You will manage the full lifecycle of machine learning models, ensuring high performance and accuracy within production environments on AWS. In this capacity, you will own the design, implementation, and continuous improvement of computer vision solutions that directly power customer identity verification. You are responsible for driving technical excellence across the team by setting standards for model development, deployment, and monitoring. The role requires deep collaboration with product and engineering stakeholders to translate business requirements into robust machine learning workflows. You will act as a technical leader in guiding best practices for data quality, model validation, and system reliability. Success in this position will be measured by the scalability, fairness, and operational stability of the biometric systems you build and maintain.

Key facts

What you'll do

• Design and build computer vision systems focused on face detection, recognition, and quality assessment, ensuring they meet stringent production requirements.
• Benchmark biometric models to ensure fairness and mitigate algorithmic bias across diverse datasets, analyzing performance disparities systematically.
• Develop and maintain end-to-end ML pipelines using Airflow for data ingestion, transformation, and cleaning to support reliable model training.
• Create synthetic data and curate balanced training sets using advanced data augmentation and collection strategies to improve model generalization.
• Optimize models for low-latency inference by applying techniques such as quantization, distillation, TensorRT, and ONNX to meet real-world deployment constraints.
• Mentor team members, perform thorough code and design reviews, and establish and enforce technical standards for machine learning development.
• Collaborate closely with cross-functional teams to define scalable architecture for biometric systems, aligning technical decisions with product goals.
• Conduct rigorous experiments to evaluate model performance, diagnosing failures and iterating on architecture and data strategies to enhance accuracy.
• Ensure that all model development practices adhere to privacy and security compliance requirements relevant to biometric data handling.
• Stay current with state-of-the-art research in computer vision and machine learning, evaluating new techniques that could provide measurable advantages in production systems.

Requirements

• 5+ years of industry experience in Machine Learning, with at least 3 years focused on Biometrics or Face Analysis, demonstrating a track record of deploying models in production.
• Proficiency in Python and vision libraries including OpenCV, Pillow, and PyTorch, with a strong understanding of underlying algorithms and data structures.
• Experience architecting end-to-end ML pipelines and using workflow orchestrators like Airflow to manage complex data and training workflows.
• Hands-on experience with cloud-native deployment on AWS, including SageMaker, EC2, and EKS, along with containerization and orchestration tools such as Docker and Kubernetes.
• Ability to scale training jobs on multi-GPU clusters, optimizing resource utilization and reducing training time effectively.
• Practical knowledge of measuring and reducing algorithmic bias in computer vision, including the use of appropriate metrics and mitigation strategies.
• Strong understanding of model evaluation metrics specific to biometric systems, such as false accept rate, false reject rate, and equal error rate.
• Experience with version control systems such as Git, along with robust experiment tracking and model registry practices to ensure reproducibility.

Nice to have

• Research publications in CVPR, ICCV, ECCV, or FG that demonstrate innovation in biometric or computer vision research.
• Experience with vector databases like Milvus or Faiss and ANN search algorithms to support efficient similarity matching at scale.
• Knowledge of privacy and security compliance for biometric systems, including frameworks relevant to identity verification and data protection.
• Familiarity with mobile or edge deployment using CoreML, LiteRT, or TFLite to enable efficient inference on resource-constrained devices.
• Experience generating synthetic training data via diffusion models or GANs to augment real-world data and improve model robustness.

Skills & tools

• PyTorch, Tensorflow, JAX
• AWS (SageMaker, EC2, EKS)
• Airflow
• Python, OpenCV, Pillow
• TensorRT, ONNX

Practical notes

The final job level for this position will be determined based on the interview process. All personal information collected during the application process is handled in accordance with the Jumio Applicant Privacy Notice.