Shalon Walter

Building the AI Medical Mentoring System (AIMMS), a full-stack medical education platform deployed at aidset.ai. The system enables medical students to practice clinical reasoning through AI-powered virtual patient simulations. Architected using React 18, TypeScript, FastAPI, and PostgreSQL. Built multimodal RAG data pipelines integrating OpenAI, ElevenLabs, Picovoice for intelligent patient interactions. Optimized pipeline performance by 40% through Redis caching and batch embedding generation.

Managed an AI summer internship program with 60 students across 12 projects, providing training in LLMs, prompt engineering, web APIs, and GenAI. Deployed multiple AI applications for academic and industry partners. Organized a final showcase with 200+ attendees attracting university and local media coverage.

Built a scalable ETL pipeline in Python with Oracle DB and Airflow integration to process 6 years of per-minute sensor data from 5,000+ environmental sensors (millions of records). Automated quality control and anomaly detection, reducing processing time by 80%. Built Tableau dashboards for real-time data quality monitoring.

Deployed an end-to-end PyTorch ML model on AWS infrastructure (EC2, S3, SageMaker) for time-series prediction using large-scale biometric data, achieving <4-day prediction accuracy in the final week before labor. Developed automated ML pipeline tools with PostgreSQL and Apache Spark.

Led research developing a U-Net deep learning model for carpal tunnel syndrome diagnosis through ultrasound image segmentation, achieving a 0.89 Dice Score. Published findings in the Medical Engineering and Physics Journal. Presented at the Orthopedic Research Society Conference.

Built enterprise integration solutions using Oracle Integration Cloud for a Fortune 500 client, connecting ERP, CRM, and data systems with custom PL/SQL logic. Designed REST/SOAP APIs and automated ETL pipelines. Improved operational efficiency by 30% through automated workflows with error handling and data validation.