Srija Sangem

Multimodal AI Interview Analysis System

- Designed a multimodal AI assessment engine leveraging real-time facial tracking, speech processing, and Large Language Models for automated candidate evaluations. - Implemented tab-switch tracking and multi-face detection features to enhance security and assessment integrity in remote interviews.

Fake Social Media Profile Detector

- Engineered a machine learning classification system to identify fraudulent accounts by analyzing user engagement metrics, profile metadata, and linguistic patterns. - Optimized ensemble model performance using stratified cross-validation and feature engineering to detect non-linear patterns of automated bot activity.