Jemima Paulraj
Data Scientist

✦ Deployed an AI forecasting model achieving 90% accuracy in predicting call volumes and staffing needs, improving customer satisfaction and reducing operational costs by 25% at Scotiabank's Peru call center.
✦ Performed in-depth EDA on 5+ years of call center data (1M+ records) from multiple domains to uncover demand patterns, seasonality, and operational bottlenecks, driving data-informed model design.
✦ Elevated the platform with an LLM-enabled experience, shifting analytics from static reports to interactive, conversational intelligence.

✦ Fine-tuned and deployed classification and object detection models on GCP (Vertex AI, AutoML Vision), achieving 95% accuracy and reducing manual labeling effort.
✦ Preprocessed 50k+ video frames for handling noise, class imbalance and annotation gaps to build high-quality training dataset.
✦ Built an end-to-end automated pipeline on GCP, processing wildlife videos from ingestion to predictions stored in BigQuery, enabling clients to showcase biodiversity and secure funding.
✦ Implemented a production-grade inference system with monitoring and retraining hooks, improving inference latency in real-world environments.
✦ Evaluated multiple YOLO variants, optimizing accuracy-latency tradeoffs to improve detection.

✦ Performed comprehensive data analysis on complex log database to generate actionable predictions and uncover critical operational patterns.
✦ Designed an interactive Tableau dashboard to visualize any unusual patterns.
✦ Proof of concept: Developed an end-to-end Android application integrating AutoML (GCP) for real-time leaf disease detection and automated remedy recommendations, streamlining field diagnostics for agricultural users.