Stephanie Muniz

Designed and deployed autonomous LLM agents using LangChain and Python for code refactoring and automated documentation; led R&D evaluating LLM reasoning using Temporal Logics (LTL/CTL) to improve reliability of code generation tools; built scalable Retrieval-Augmented Generation (RAG) pipelines linking enterprise knowledge bases with conversational AI, integrating logic via FastAPI. Improved LLM service performance through prompt optimization, caching, batching, and inference tuning, reducing average response latency by 35% for production AI workflows. Tuned vector DB retrieval layers (Pinecone/Milvus/Weaviate) for enhanced semantic search quality and metadata filtering. Led Azure-based experimentation, model validation, and deployment workflows for enterprise GenAI solutions. Collaborated with clients to define GenAI use cases and shape technical roadmaps.

Specialized in Customer Experience AI, NLP automation, and scalable ML deployment. Developed sentiment analysis models (LSTMs and Transformers) for voice and text channels; engineered chatbot flows to automate debt negotiations; integrated ML predictions into the core CX platform, ensuring low-latency responses. Strengthened system robustness with rigorous unit testing and automated validation pipelines. Deployed NLP models on AWS (SageMaker, EC2, Lambda) for scalable inference in CX applications.

Designed supervised models for forecasting and anomaly detection in finance and operations; built RESTful APIs to expose ML predictions; conducted Exploratory Data Analysis and data visualization; managed data pipelines using Python, Pandas, PostgreSQL, and SQL Server; developed React-based dashboards to present insights to stakeholders.