HARSH ANANT

Results-driven developer with over 3+ years of experience in delivering projects for insurance client. Expertise in developing python scripts and ETL jobs ensuring successful project delivery. Good understanding of machine learning and deep learning algorithms and hands on experience in implementing them using tensorflow, keras and scikit-learn. Proficient in developing, training, feature engineering, model evaluation (cross-validation, accuracy metrics), and optimizing performance through hyper parameter tuning and regularization techniques for predictive analysis, classification and regression tasks.

Compare the performance of different machine learning and deep learning algorithms on the data set

• Compared performance of different ML classifiers such as Logistic Regression, KNN, Decision Tree and Random Forest on one multiple class data set, and different fully connected Deep Neural Network architectures with optimizers such as SGD, Adam and RMSprop and also with dropout for any one of the architectures on other multi class data set with respect to recall, precision and f1 score.
• Decision Tree achieved 99.9% accuracy in testing, as well as in training
• Deep neural network with Adam optimizer achieved 52% accuracy in 12 epochs of training and validation

Design the Convolutional Neural Network and Test the Accuracy on the Data Set

• Read the research paper on ‘Wild Animal Classifier Using CNN’ from ‘ResearchGate’.
• I took the wildlife data set from Kaggle, followed the steps of data pre-processing, CNN model development, and training
• Model achieved 95% training accuracy, and 92% testing accuracy

Microsoft Certified: Azure Data Scientist Associate