Jayasurya Vijayakumar

• Developed and optimized credit risk models using logistic regression, XGBoost, and other machine learning techniques, iterating through various approaches to enhance predictive accuracy. Implemented effective feature selection and extensive bureau feature engineering to improve model performance and risk assessment.
• Utilized LangChain to classify customer call transcripts, identify intent, summarize conversations, and resolve customer queries in the banking sector.
• Developed and optimized functions on the Leaps platform, conducting performance and load testing to ensure optimal performance. Identified and resolved issues within functions to enhance functionality and efficiency.
• Collected data related to Aadhar and PAN cards to develop a transfer learning model for an image classification project. Leveraged transfer learning techniques to identify Aadhar and PAN cards accurately, contributing to improved image classification capabilities.
• Implemented face duplication detection for identity verification using ArcFace and FaceNet models, enhancing the reliability and accuracy of facial recognition systems.

• Developed sentiment analysis models using both machine learning and deep learning techniques, capable of analyzing sentiments and aspects of sentiments across multiple languages.

Implemented a language translation feature in the sentiment analysis model to convert non-English texts to English for comprehensive sentiment analysis.

• Created various survey analyses, including heatmaps, text highlighters, max diff, and conjoint analysis, providing insights into user preferences and behavior. Utilized Python to analyze user behavior and identify patterns in survey responses, enabling data-driven decision-making.

• Developed an Aadhar masking solution using OCR (Optical Character Recognition) techniques to detect Aadhar images uploaded by users and mask the Aadhar numbers, ensuring data privacy and security. Conducted comprehensive data collection from diverse sources, including e-commerce platforms and online platforms, leveraging web scraping techniques to gather relevant data for sentiment analysis model development.