Data Scientist, Computer Vision

Overview

BWXT Advanced Technologies is seeking a Data Scientist, Computer Vision (Classification & Deep Learning) to design, train, evaluate, and product ionize image classification models that power critical decisions across our products and operations. You will own datasets, modeling, and deployment for robust, scalable visual classification—delivering measurable accuracy, reliability, and latency improvements.

This position is based on-site in Lynchburg, VA at the Advanced Technologies Office.

• Lead end to end Computer Vision classification: problem definition, dataset creation, experiment design, model training, evaluation, deployment, and monitoring.
• Develop modern deep learning models using CNNs (Res Net/Efficient Net) and Vision Transformers (ViT/Swin), leveraging transfer learning, fine tuning, and self /weakly supervised methods as appropriate.
• Handle imbalanced/noisy/multi label data with class aware sampling, focal/cost sensitive losses, label smoothing, and advanced augmentations (Rand Augment, Mix Up, Cut Mix).
• Establish rigorous evaluation: precision/recall, F1, ROC/PR AUC, calibration, confusion analysis, per class metrics, subgroup fairness, and stress testing for lighting, occlusion, motion blur, and device variation.
• Build data & experiment pipelines: image ingestion, labeling QC, dataset versioning and experiment tracking with automated reproducibility.
• Production operations: deploy services, implement drift detection and alerting, schedule retraining, support A/B tests and human in the loop review.
• Cross functional collaboration with data engineering, product, operations, and quality to integrate outputs into workflows and dashboards.

• A bachelor’s degree in computer science, electrical engineering, physics, or related field is required.
• A minimum of six (6) years of building and deploying computer vision classification models in production or related work experience is required.
• Must have strong experience with PyTorch (preferred) or Tensor Flow/Keras; along with proficiency in Python (Num Py/Pandas); and familiarity with scikit learn for baselines and metrics.
• Must have hands on with OpenCV, torch vision/timm, albumentations; image pre /post processing and dataset curation.
• Must have demonstrated expertise with CNNs & Vision Transformers, transfer/self supervised learning (e.g., SimCLR/MoCo/DINO/MAE), mixed precision training, and training efficiency.
• Must have experience exporting and serving models (ONNX, Tensor
  
  RT/OpenVINO), containerization (Docker), and CI/CD for ML services.
• Must be able to communicate effectively and translate model results into actionable product/operations insights.
• Must have a deep understanding of image classification theory and practice: loss functions, optimization, augmentation, calibration, and thresholding.
• Must have strong software engineering discipline: code reviews, testing, logging/observability.
• Must be proficient in experiment design, statistical analysis, and scientific communication.
• Must have a strong understanding of security/privacy best practices for visual data (PII/PHI as applicable).
• Must be a U.S. citizen.
• Must be able to obtain and maintain a U.S. Department of Energy (DOE) or Department of Defense (DOD) security clearance, whichever is required.

• MS/PhD in Computer Science, Electrical Engineering, Physics, or related field.
• Edge inference (NVIDIA Jetson/ARM), streaming pipelines, or multi camera systems.
• Data labeling operations (CVAT/Label Studio), quality control, and consensus strategies.
• Robustness to domain shift; techniques for generalization across environments/devices.
• Weak supervision, active learning, or semi-automated data curation.
• Interpretability (Grad CAM), calibration, and documentation (model cards, datasheets).

• Competitive salary and benefits package, including health, dental, and retirement plans.
• Flexible work schedules and paid time off to promote a healthy work-life balance.
• Professional development opportunities, including mentorship programs and sponsorship for continuing education.
• An inclusive atmosphere…