Prashant Chowdhary

• Pioneered the development and utilization of advanced NLP models utilizing deep learning techniques like LSTM, BERT, and Transformers for accurate classification of unstructured text.
• Employed eli5, shap, and limeTextExplainer for improved model interpretability.
• Enhanced operational efficiency by implementing automated infrastructure tasks with the use of Terraform and AWS Lambda.
• Completed Quantexa training with a focus on dynamic entity resolution and network generation.

• Extracted and transformed large datasets from Enterprise Data Warehouse to Data Lake.
• Built Feature Bank for machine learning models, enhancing prediction accuracy.
• The successful implementation of an optimized AML Transaction Monitoring model was achieved through my expertise in XGBoost and hyper-parameter optimization techniques using RandomizedCV and GridsearchCV.
• Expertly tracked model performance and seamlessly integrated model recommendations into operational systems.

Enrichment Features Optimization Model:

• Achieved significant time savings through accurate prediction of optimal enrichment variables, resulting in a 40% reduction in costs.

Attribution Model:

• Utilized Click Decay and Markov Chains to identify the key elements influencing conversions.
• Utilized findings to optimize similar product categories based on customer journeys.

Markov Chain Model:

• Mapped and analyzed customer journeys using probabilistic modeling techniques to gain insights into application interactions.
• Transition probabilities were estimated to determine the conversions on particular channels.