Harshit Garg Email & Phone Number

Charlotte, NC, US

Johns Creek, Georgia, United States

• Collaborated closely with Research Scientists to build a bespoke, Large Language Model-powered tool, aligning generative AI capabilities with key use cases in lens design.
• Architected a scalable Retrieval Augmented Generation (RAG) pipeline on AWS SageMaker, integrating Meta's LLaMA 3.1 model. Led end-to-end implementation, including data retrieval, chunking, embedding, and storage in.
• Deployed models using AWS SageMaker, Lambda, and integrated Amazon Lex for UI, allowing real-time interaction and feedback. Additionally, stored feedback in DynamoDB to refine the model's performance based on user.
• Utilized GraphRAGs to enhance model performance, improving the retrieval and contextualization of information by integrating knowledge graphs with retrieval-augmented generation.
• Explored cutting-edge LLM technologies, experimenting with knowledge bases on Amazon Bedrock and integrating Anthropic Claude 2.1 to refine the AI pipeline for better accuracy.
• Expanded cloud platform expertise by incorporating Azure AI Studio and GPT-4 into the workflow, developing a cross-platform, cloud-agnostic solution.

Bengaluru, Karnataka, India

• Led a team of 4 to build and deliver major functionalities in the business lending application, coordinating development efforts, client interactions, and project timelines.
• Managed Git repositories and implemented CI/CD pipelines using Jenkins, enhancing deployment efficiency and version control practices.
• Developed and deployed a Fire and Smoke Detection system using various machine learning models (YOLOv7, YOLOv8, YOLOX, EfficientDet, FasterRCNN), achieving a 96.7% accuracy rate with YOLOX.
• Built a drone imagery solution for detecting damaged electric poles post-storm, achieving 92.39% accuracy using advanced ML models.
• Designed a parking lot management system utilizing OCR for license plate recognition, integrated with an Angular-based dashboard for real-time tracking.
• Worked on X-ray scanner image analysis to detect harmful objects using the Segment Anything Model (SAM), showcasing advanced image processing and ML techniques.

Bengaluru, Karnataka, India

• Developed full-stack functionalities for a U.S. bank’s business lending application using Spring Boot microservices, PLSQL, and Angular.
• Collaborated directly with clients to gather requirements and understand use cases, ensuring seamless delivery of requested features.
• Contributed to the design and development of critical application components to enhance user experience and streamline backend processes.

Master'S Degree, Data Science

Courses: 1. MATH677 - Mathematical Foundations for Data Science 2. STAT650 - Statistical Foundation For Data Science 3. STAT624.

Bachelor Of Technology - Btech, Computer Science Engineering, 9.1/10.0