FATEMA BOHRI

Client: Verizon Sourcing LLC

• Reduced inquiry resolution time by 96% (2 hours to 5 minutes) by designing and deploying a production-grade Agentic AI system that automatically classified and routed requests to specialized agents for Verizon's OM and account teams.
• Automated recurring status workflows, recovering 14+ engineering hours per month and reducing operational escalations by 80%, allowing teams to focus on high-value operational work.
• Deployed the solution on Verizon's Open Agent Platform (OAP) and Enterprise GenAI and Agentic AI Platform (VEGAS), wiring it into internal APIs (IEN, SR, Milestone View) and a secure Node.js Slack middleware (Slack Bolt + Socket Mode) routed through the corporate proxy.
• Stack: LangGraph · LangChain · Python · Node.js · VectorDB · GitLab

• Developed an end-to-end machine learning solution for vehicle mileage prediction, achieving 80% accuracy through feature engineering, model optimization, and rigorous performance evaluation.
• Engineered the complete application stack, integrating a Java-based inference backend with a responsive web interface to deliver real-time predictions and a seamless user experience.