My Projects

Agentic AI systems I've built, from clinical assistants to enterprise document Q&A

Securing Agents - Multi-Agent Customer Support

Course Project | UT Austin McCombs - AI Agents for Business Applications | 2026

A LangGraph-orchestrated customer support assistant with production-style security guardrails: prompt-injection detection, output-safety scanning, tier-based access control, and a full audit trail across four specialist support agents.

LangGraphAI SecurityMulti-Agent SystemsStreamlitGuardrails

Last-Mile Delivery Exception Agents

Course Project | UT Austin McCombs - AI Agents for Business Applications | 2026

A LangGraph multi-agent system that triages and resolves last-mile delivery exceptions - deduplicating noisy shipment logs, deciding a resolution against an operational playbook via RAG, escalating to a human when policy requires it, and drafting a personalized customer notification, with a full auditable decision trail.

LangGraphMulti-Agent SystemsRAGPythonHuman-in-the-loop

AI-Powered EHR Assistant

Capstone Project | UT Austin McCombs - AI Agents for Business Applications | 2026

An agentic assistant that helps clinicians navigate electronic health records - retrieving patient history, summarizing chart notes, and answering natural-language questions grounded in structured and unstructured EHR data, with human-in-the-loop safeguards for clinical accuracy.

LangChainAgent OrchestrationPrompt EngineeringPythonHuman-in-the-loop

RAG System for Enterprise Document Q&A

Capstone Project | UT Austin McCombs - AI Agents for Business Applications | 2026

A Retrieval-Augmented Generation system that lets users ask natural-language questions over a long-form Harvard Business Review case study on Apple, combining vector search with an LLM to ground answers in the source document and reduce hallucination.

RAGVector SearchLangChainModel Context Protocol (MCP)Evals