Machine Learning Scientist | Deep Learning, Mlops, Predictive Analytics
Current• Streamlined interviews with business partners to redefine the ‘Car and flows distribution issue,’ which led to the development of a machine learning model and performance metrics designed to address operational realities. • Conducted a comprehensive assessment of machine learning algorithms (SVM, ARIMA, KNN) and deep learning models (CNN) to determine the most effective approach, leading to a 20% improvement in revenue forecasting accuracy. • Pioneered the development and application of a cutting-edge deep learning model, fusing 1D CNN with LSTM, resulting in a remarkable reduction in RMSE by one to two orders of magnitude and outperforming the accuracy of the open-source model by approximately 45%.• Developed and launched a robust MLOps dashboard to continuously monitor model performance and input and output metrics. Streamlined concept drift and data drift detection, resulting in a 40% reduction in model retraining time.• Proposed an innovative architecture combining M-GCN and LSTM models, enhanced with an attention mechanism that could revolutionize traffic management approaches.
