Tanay Mevada

About Tanay Mevada

Data Engineer with over 5 years of experience building scalable cloud pipelines, Document…

Experience

1. Data Scientist

   Verizon

   Jun 2023 — Present

   Developed a scalable document classification pipeline to automatically organize 50K+ unstructured documents, reducing manual review efforts and improving accessibility for multiple internal teams-Used an Isolation Forest model in the first stage to detect and filter out-of-class or noisy documents, improving the quality of training data for subsequent models by ~10%-Trained a Random Forest classifier using TF-IDF embeddings to group documents into six main categories (e.g, patient notes, billing notes), achieving around 90% accuracy on validation data-Built and optimized class-specific Hierarchical Attention Network (HAN) models to handle more nuanced, within-category distinctions, which helped improve classification consistency and interpretability-Combined TF-IDF and BERT embeddings at different stages to balance performance with efficiency, while performing feature engineering, selection, and evaluation to ensure stable model behavior-Deployed the models on AWS SageMaker and set up CI/CD pipelines for retraining and updates, reducing manual intervention and deployment time by ~40%-Monitored model drift and concept drift over time using scheduled evaluation jobs, ensuring the models continued to perform reliably as new document types were added.

Education

• Gujarat Technological University, Ahmedbabd

  Bachelor's degree

  2018 — 2022

• Saint Peter's University

  Masters

  2022 — 2024