Projects Profile Company: Verizon Role: Data Scientist Oct ,2019- Present Work Location Hyderabad, India Technologies Machine Learning, Deep Learning, NLP, Python, Tensorflow, Keras, OpenCV, Pyspark. SQL, Hive, Statistics, Tableau, R, Pytorch Projects and Responsibilities: Image Labeling and Crack Detection Model :
• Built image labeling and Crack detection Model to automate the device Exchange/Upgrade process for the customer.
• Used OpenCv for image preprocessing and CNN, Transfer Learning technique to implement the model. Used different optimization techniques to improve the model performance.
• Worked with Data Engineering teams to deploy the model and made all required changes to automate the process in the production environment. Abstractive Summarization Model :
• Developed Abstractive Summarization model to generate human likes summaries without missing any keys context from the call transcript. • Used Deep learning and Google Pegasus technique to implement the model in production.
• Implemented TF-IDF,Named Entity Recognition, POS tagging, N-gram and Word2Vec in Natural Language Processing model to improve model performance. Auto Tagging Model :
• Implemented a model to provide the tags of Customer feedback to understand the pain area of customers. Used Multilabel Classification to meet project milestones.
• Cleaned the text data and used different word embedding technique in model to improve model performance.
• Deployed the model in develop environment to automate the process. Cyber Attack Detection Model:
• Built a Traffic Detection model using the Deep learning technique to predict the cyber attack.
• Collected the Data from Database with help of SQL and Explored the data analysis for internal business performance.
• Built the model where Variance-Bias trade-off is taken care, which results in less error with high accuracy. Recommendation Model:
• Developed recommendation model to help identify the loyal Customer and predict the likelihood of Customer buying a recommended products
• Performed Multiclass classification, Collaborative, Content-Based, Cold Start and Matrix factorization based methods using Pyspark to build the Model. Used Hive for collecting the whole data from database
• Conducted different statistical tests to optimizes features collection and made the model for business outcome