Associate Software Engineer, T-Mobile - Chicago, IL

03/2025 to Present

• Re-engineered agent tooling and evaluation loops for an internal GenAI copilot by implementing LLM-as-a-judge validation with correctness and tool-output verification, increasing response reliability by 48%.
• Built a multi-agent orchestration framework using LangGraph and MCP-style structured contracts and guardrails, improving tool-call accuracy by 160% and enabling safer multi-step reasoning workflows in production.
• Designed and shipped quality and efficiency optimizations including prompt/context engineering, dynamic summarization, and token budgeting, cutting LLM cost by 70% while maintaining output quality.
• Developed product-facing personalization and UI tooling in Streamlit including context-aware suggested prompts and role-based behavior, improving usability and reducing time-to-resolution.
• Partnered cross-functionally to define measurable quality goals, debug complex failures, and ship reliable AI features under production constraints.