Srujan Pendurthi

Louisville, KY

• Streamlined Medical Assistant workflows by 33% by developing multimodal RAG pipelines using Google ADK.
• Consolidated patient data and charts into Athena EMR using FastAPI and Snowflake MCP.
• Increased patient throughput by 6% by implementing LangChain and LangGraph agents into the MA workflow.
• Improved veteran healthcare data integration by creating API endpoints for Humana Military TRICARE.
• Presented in the final showcase to senior leadership and stakeholders.

• Improved personalized learning recommendations by 28% by developing NLP and computer vision models in PyTorch for AI-driven e-learning tools.
• Reduced data processing time by 35% by building analytics workflows with Redis and NumPy on educational data.
• Increased platform engagement by 18% by helping deploy and optimize scalable end-to-end ML inference pipelines.

Phoenix, AZ

• Developed a real-time ASL gesture translation system using OpenCV, MediaPipe, and PyTorch.
• Applied Gaussian blur and thresholding techniques to improve hand segmentation and gesture feature extraction.
• Integrated EEG signal data with gesture recognition pipelines for BCI-focused multimodal classification research.
• Improved sign recognition accuracy by 25% by training an RNN classifier on temporal gesture sequence data.
• Authored a 7-page research paper and presented findings at the Arizona Association of the Deaf State Conference.