Prajith Pandrate Email & Phone Number

Washington, DC, US

• Designed and optimized a LLM compression algorithm with CUDA kernels. Cut model size by 35% and improved inference speed by 20%, while maintaining accuracy with perplexity and F1 score evaluations.
• Built and deployed interactive data visualization tools using Numpy, Matplotlib, and Scikit-learn. Integrated CI/CD pipelines to automate deployments, which reduced downtime and efficient updates.

• Refactored data pipelines using Python, SQL, and AWS, which reduced data processing times by 20%, and utilized Linux/Ubuntu environments to handle high-volume research datasets more efficiently.
• Led the migration of the Wukong parallel computing framework to Jupyter Notebook. Leveraged AWS Lambda, IAM roles, and S3 to improve research collaboration across teams.

• Developed a Dialogflow-based chatbot with Python and JavaScript to improve healthcare access for over 500 users in rural India. Leveraged natural language processing (NLP) to enhance user interaction and automate.
• Conducted in-depth health market analysis and utilized SQL and MATLAB to analyze demographic data. Utilized data to design of a tailored solution for underserved populations.

• Improved platform by using RESTful APIs to enhance system scalability. The improved data accuracy and responsiveness resulted in a 25% increase in user engagement.
• Built a health data collection application using FHIR (Fast Healthcare Interoperability Resources) standards and MySQL for robust data management. Ensured real-time user support through a React.js-based conversational.

Computer Science And Economics

Education record

Activities and Societies: VEX Robotics, Debate (President), Science Olympiad (President) Attended the Academy of Science at the Academies.