Open to opportunities

Donnel Celwin

@donnelcelwin

Message

Senior AI/ML Engineer building production healthcare ML systems and MLOps for fast, responsible model delivery.

United States

Message

What I'm looking for

I’m looking to lead production AI/ML and MLOps—building scalable inference, fast refresh cycles, and responsible AI—especially in healthcare use cases. I want teams that value measurable impact, strong ML foundations, and rigorous monitoring.

I’m a Senior AI/ML Engineer with eight years building production machine learning systems, with a strong focus on clinical risk prediction, NLP, and demand forecasting. I thrive on turning experimentation into reliable, continuously improving services that teams can trust.

At Dreamlabsdigital, I lead the ML platform powering 12 production models across healthcare risk scoring, demand forecasting, and NLP document processing. I’ve driven an average model refresh cycle of 48 hours instead of the industry-typical 6 weeks, and delivered measurable outcomes like 0.89 AUC ROC for 30 day readmission prediction and a 14% reduction in readmission rates for hospital partners.

I bring deep MLOps experience with AWS SageMaker, Kubernetes, and MLflow—owning the full lifecycle from experimentation to serving. I’ve built real-time inference infrastructure on Kubernetes (autoscaling GPU nodes, serving 2.8M predictions daily at p99 latency under 120ms) and embedded responsible AI practices like bias auditing, SHAP-based explainability, and data drift monitoring to catch silent failures before they impact clinical decision support.

Experience

Work history, roles, and key accomplishments

Current

Senior AI/ML Engineer

Current

Dreamlabsdigital

Apr 2022 - Present (4 years 1 month)

Led the ML engineering team of 4 to build and operate 12 production models across healthcare risk prediction, demand forecasting, and NLP document processing. Cut average model refresh time from 6 weeks to 48 hours and delivered clinical risk models with 0.89 AUC ROC plus 14% readmission reduction for partner hospitals.

AWS Sagemaker MLFlow Kubernetes XGBoost LSTM Ensembles BERT SHAP

ML Engineer

Tempus AI

Aug 2019 - Mar 2022 (2 years 7 months)

Built production ML pipelines for precision oncology and clinical NLP, including a transformer-based clinical trial matching engine with 91% precision processing 12k new profiles monthly. Engineered feature stores and monitoring (Feast, Delta Lake, MLflow) to consolidate features across 8 sources, reduce duplication by 70%, and enable proactive retraining ahead of SLA-impacting performance drops.

PyTorch Feast Delta Lake MLFlow Airflow Spark SQL Model Monitoring

Data Scientist

Uptake Technologies

Jun 2017 - Jul 2019 (2 years 1 month)

Developed predictive maintenance and anomaly detection models for industrial IoT, processing 500M daily sensor readings to forecast equipment failures 7–14 days in advance. Delivered production models for 4 enterprise clients generating an estimated $12M in avoided downtime annually, using isolation forests and autoencoders with 92% true positive rate and under 2% false alarms.

Predictive Maintenance Anomaly Detection Isolation Forest Spark Industrial IoT Sensor Data