Mohansai S

Sr Data Scientist @Vention

US

Built predictive algorithms using XGBoost and Prophet to forecast machine downtime with 92% accuracy, increasing production efficiency by 21%. Developed end-to-end demand forecasting system using LSTM neural networks, enhancing inventory planning and reducing deadstock by 27%. Created anomaly detection pipeline using unsupervised clustering (DBSCAN, Isolation Forest) for real-time monitoring of robotic processes. Integrated ML models into edge devices through Dockerized microservices, reducing latency by 40%. Constructed custom ETL workflows with Spark and AWS Glue to handle multi-source IoT sensor data. Partnered with operations team to develop ML models for dynamic supply chain adjustments, cutting logistics delays by 18%. Built interactive Tableau dashboards linked with real-time ML outputs, enabling instant executive decision-making. Used SHAP and LIME to build interpretability modules, improving stakeholder trust in ML results. Published weekly model performance reports and presented KPI impact to leadership using SQL and Power BI insights. Conducted advanced root cause analysis using SHAP-based interpretability layers in deployed models, directly influencing 3+ engineering change decisions and enhancing model trust across technical and executive teams. Conducted model drift monitoring using AWS SageMaker Model Monitor and automated retraining strategies. Improved model training pipeline efficiency by 36% using parallel processing and GPU acceleration. Integrated predictive quality control models using computer vision (OpenCV + PyTorch) to identify surface defects on assembly line components, reducing defect rate by 26% and improving first-pass yield.

Narayana Junior College - India

Intermediate

Narayana EM School

Secondary School Education

University of Delaware

Master of Science - MS

Sree Vidyanikethan Engineering College

Bachelor of Technology - BTech