Madhushan Buwaneswaran

WORK HISTORY

EDUCATION

ABOUT MADHUSHAN BUWANESWARAN

Machine Learning and Data Science Engineer with over 3 years of industry experience and more than 2 years of research experience. I have developed and deployed scalable machine learning systems optimized for real-world applications in the telecommunication and middleware domains.Technical Skills & Tools: Python, SQL, R, PyTorch, Scikit-learn, TensorFlow, Numpy, Pandas, Git/GitHub, Azure MLSome highlights of my work include:• Designed, developed, and deployed a real-time face recognition system with 98% accuracy for employee and customer identification for a Tier-1 Telco provider. The FR system includes state-of-the-art face detection, alignment, and recognition models together with face tracking and multi-frame validation to reduce false positives.• Developed an unsupervised multivariate time-series anomaly detection model to identify performance anomalies in microservices hosted on a cloud PaaS with the aim of reducing service downtime. Deployed the model using ONNX and Azure ML and integrated it with Azure Streaming Analytics for near real-time detection and alerting.• Researched and developed a machine learning approach for imputing missing data in smart meter data in collaboration with a leading meter data management company in North America. The approach is designed to operate with low data and computation resources, making it suitable for real-world applications.My research interests are focused on improving the data and computational efficiency of deep-learning models in computer vision and time series domains. I am also interested in statistics and its insights. I share my thoughts on the topics I am interested in, at bmshan.medium.com.I am interested in opportunities where I can apply machine learning and data science concepts to solve real-world problems, resulting in tangible positive impacts.