Ronny Sarris

Designed and deployed a voice-enabled AI assistant and chatbot for medical staff, providing real-time decision support and personalized care recommendations. Built MCP-based LLM integrations, RAG and multi-agent architectures, and LLMOps/CI-CD pipelines to ship production systems on Azure using LangChain/LangGraph and fine-tuned LLaMA (LoRA/QLoRA).

Engineered conversational AI and RAG pipelines for healthcare payers and providers to improve care quality while streamlining operations and reducing administrative costs. Implemented HIPAA-aligned role/policy access controls, care gap detection, and claims-management automation, and built React dashboards for monitoring and evaluation metrics.

Developed TensorFlow NLP models to extract sentiments and product features from millions of customer reviews powering review highlight experiences. Deployed models on AWS (SageMaker, Lambda, Redshift) with offline evaluation and A/B testing, and built analytics dashboards in QuickSight and Tableau.

Conducted applied research in anomaly detection and unsupervised learning to identify financial fraud, cyber threats, and operational risks. Built and evaluated models using clustering, autoencoders, Bayesian methods, and graph analytics, and scaled Python prototypes into production-ready systems.

Built and maintained Python ETL pipelines for large-scale image datasets to support computer vision model training and production deployments. Optimized ingestion with PostgreSQL, Hadoop, and AWS S3, automated dataset validation/anomaly detection (reducing data errors by 30%), and developed internal data APIs and monitoring for experiment reproducibility.

Activities and societies: Research Assistant; Teaching Assistant; Object Oriented Programming

Earned a Bachelor of Computer Science at Georgia State University, with experience as a research assistant and teaching assistant supporting coursework in object-oriented programming.