- Optimized Operational Costs: Automated Time Series Forecasting of Agent Headcount, ensuring SLA compliance and cost-efficient staffing.- Enhanced Predictability: Modeled BandWidth Requirements using Salesforce data and historical traffic patterns to ensure reliable traffic management.- Improved Platform Performance: Developed dashboards and data models tracking key KPIs, leading cross-functional teams to enhance user journeys and service reliability.

- Adaptive Practice Enhancement: Developed an XGBoost Regression Model and a Spark Pipeline to tag and calibrate question difficulty based on student attempts.- Automated Tagging: Fine-tuned Large Language Models on an academic corpus to create a classifier recommending concepts from 60,000+ Knowledge Graph Nodes, reducing manual tagging effort.- Predictive Modeling: Deployed an LSTM Model (DKT) to predict student correctness probability, using Knowledge Graph Embeddings to improve interpretability.- Dashboard and Notifications: Led the Growth Team in creating dashboards with Data Engineering and partnering with Marketing for effective notification delivery.

Question Difficulty Level Prediction ( Supervised Learning )