• Designed and deployed production ML pipelines for healthcare analytics using Python, Pandas, NumPy, and scikit-learn, improving model reliability and downstream data quality by 28%.

• Built and fine-tuned transformer-based NLP models for clinical text classification and entity extraction using BERT architectures, reducing manual review effort for research teams by 35%.

• Implemented end-to-end MLOps workflows using Docker, MLflow, and GitHub Actions, standardizing model versioning, experiment tracking, and CI pipelines across teams.

• Deployed real-time inference services using FastAPI and REST APIs on AWS EC2 and S3, enabling scalable model serving with consistent latency under production workloads (22% latency reduction).

• Collaborated with data science, regulatory, and product stakeholders to translate clinical and business requirements into deployable ML solutions, ensuring auditability and compliance readiness.

• Integrated monitoring and alerting using Prometheus and custom logging to track data drift, model performance, and service health, improving incident detection without disrupting clinical operations.

• Developed supervised and unsupervised ML models for predictive analytics and recommender systems using Python, scikit-learn, XGBoost, and LightGBM, increasing client KPI accuracy by 31%.

• Designed time-series forecasting solutions using statistical models and deep learning approaches, enabling demand and trend prediction that improved planning decisions by 24%.

• Built NLP pipelines for sentiment analysis, text classification, and named entity recognition using transformers and traditional NLP techniques, supporting data-driven product insights for multiple clients.

• Implemented computer vision models for image classification and object detection using CNN architectures, delivering automated quality checks and visual analysis workflows.

• Deployed ML models as scalable APIs using Flask and FastAPI, integrating with client applications through REST interfaces and ensuring production readiness.

• Established model experimentation, tracking, and reproducibility practices using MLflow and structured Git workflows, reducing rework and model regression issues by 27%.

• Partnered with cross-functional teams to perform data validation, feature engineering, and exploratory analysis using SQL, Pandas, and visualization tools, translating raw data into actionable insights.