Machine Learning Engineer - Computer Vision & GenAI

Nora Alaoui Machine Learning Engineer Page * of *

Nora Alaoui, M.S.

MACHINE LEARNING ENGINEER AI ENGINEER COMPUTER VISION ****@*******.*** 202-***-****

github.com/noraleilaa linkedin.com/in/nora-alaoui-b49968135 nalaoui.com PROFESSIONAL SUMMARY

Machine Learning Engineer with 4+ years of experience building computer vision, clinical ML, and generative AI systems. Owns Python and PyTorch workflows from data preparation and training through evaluation, deployment, and monitoring. Recent work includes rare-event prediction, patient graphs, RAG, protein generation, and YOLO/Detectron2.

TECHNICAL SKILLS

Data Science: Python, SQL, R, PySpark, scikit-learn, XGBoost, pandas, NumPy, feature engineering, time-based validation

Graph ML: trained graph neural networks (GNNs), GraphSAGE Computer Vision: YOLO, Detectron2, OpenCV, CNNs, ResNet, auto-labeling, semi-supervised learning, 3D Slicer, edge inference

Clinical Data: EHR and ECG data, MUSE XML, AUROC, AUPRC, calibration, SHAP, fairness Generative AI / MLOps: transformers, RAG, diffusion models, protein generation, Docker, FastAPI, AWS SageMaker, CI/CD, inference APIs, model monitoring EXPERIENCE

Cardiology Deep Learning Research Trainee Children's National Medical Center May 2026 - Present

• Built an end-to-end arrhythmia prediction pipeline for 4,918 cardiac-surgery patients and 34,204 patient-time records, combining demographics, diagnoses, procedures, medications, labs, and patient history.

• Developed preoperative and rolling postoperative models with logistic regression and XGBoost. Used time- based splits and only data available at prediction time to prevent data leakage.

• Built patient graphs and similarity features for graph-based modeling. The best model reached 0.748 test AUROC and 3.3x baseline AUPRC on a highly imbalanced outcome.

• Prepared 12-lead ECG and MUSE XML data and tested CNN models for use with the clinical risk pipeline. Machine Learning & Generative AI Engineer ProSyn Jun 2024 - Present

• Designed a RAG-based multimodal pipeline that combined biomedical literature retrieval, transformer models, and protein-generation models to produce research candidates and comparison reports.

• Deployed containerized Docker and FastAPI inference services and built AWS SageMaker training workflows for PyTorch transformer and diffusion models.

• Built protein comparison and review tools that displayed model confidence, structure differences, and source evidence; saved expert feedback for later model runs.

• Optimized GPU training and inference, reducing protein-generation time by 40% and improving training stability by 20%.

Nora Alaoui Machine Learning Engineer Page 2 of 2 EXPERIENCE CONTINUED

Machine Learning / AI Researcher & Developer Virginia Tech Arlington Innovation Center Dec 2021 - Jun 2022

• Built TensorFlow CNN and transfer-learning models for lung-cancer images, reaching 80% test accuracy; used 3D Slicer augmentation to reduce training time by 30%.

• Trained YOLO and Detectron2 models to find lung lesions and compared accuracy, recall, false positives, and speed. Used model-generated draft labels and expert review to add more labeled CT images.

• Built a repeatable imaging-data pipeline for Cancer Imaging Archive datasets, including quality checks, preprocessing, augmentation, and held-out model testing. Machine Learning Scientist mdlogix Apr 2021 - May 2022

• Built mental-health models that predicted 80+ suicide events and 130+ adolescent crises, giving care teams earlier data for review and follow-up.

• Moved production analytics from AWS to Oracle and automated ETL pipelines, improving processing speed by about 25% and reducing manual data preparation by 40%.

• Added HIPAA-aware data checks and repeatable model workflows, improving data quality and making training runs easier to reproduce.

SELECTED ML & AI PROJECTS

Real-Time Object Detection for Autonomous Driving Northwestern University, 2024: Led a real-time object detection pipeline for edge devices; matched the model to GPU and NPU limits and tracked model accuracy and data quality.

Facial Recognition in Augmented Reality Northwestern University, 2024: Built a 26-layer ResNet facial- recognition model and reached 90% accuracy while meeting real-time speed goals. Object Detection App Northwestern University, 2024: Built an iOS app that used cloud vision and RAG to identify objects and return simple context to the user. RESEARCH & PRESENTATIONS

Bridging the Translation Chasm with Automated Machine Learning Conference on AI, Georgetown University, 2021: Presented mdlogix research on moving machine-learning results into mental-health workflows. Combat-Related PTSD and Virtual Reality Treatment Abertay University, 2020: Presented emerging virtual- reality approaches for PTSD treatment.

EDUCATION

Northwestern University, Evanston, IL - M.S., Data Science (Artificial Intelligence), 2024 Selected study: Computer Vision, Deep Learning, Practical Machine Learning, Generative AI, NLP, Data Engineering, and AI Ethics

George Mason University, Fairfax, VA - B.S., Psychology (Neuroscience) Selected study: Neuroscience, Human-Computer Interaction, Technology in Mental Health, and Virtual Reality

Contact this candidate