SWETANSHU AGRAWAL

• Collaborated with a team of developers to identify and resolve challenges in the application development process, enhancing overall workflow and efficiency.
• Designed and implemented deep learning models to increase the accuracy and efficiency of existing AI applications, focusing on practical improvements and outcomes.
• Utilized advanced hyperparameter tuning techniques to optimize neural networks, achieving significant enhancements in model performance.

Additional Project Description:

• Developed a deep learning model for object detection, successfully converting the code into lightweight and efficient formats compatible with ESP32 and Arduino NAO platforms. This adaptation involved using both TensorFlow Lite and ONNX frameworks to ensure broad compatibility and performance efficiency on resource-constrained devices.

• Developed and implemented a Library Management System using Swift, SwiftUI, and Firebase, designed to efficiently track records of books, fines, members, and other related data. Ensured system configuration adhered to best practices and standards for usability and data integrity.
• Addressed non-functional requirements, including system performance, security, and maintainability, to ensure robust and reliable operations.
• Created and maintained iOS applications, leveraging skills in Objective-C, Swift, and Xcode to build user-friendly and performant apps.

• Efficiently organized and prioritized tasks to achieve service goals, while continuously updating industry knowledge through advanced training.
• Successfully integrated Natural Language Processing (NLP) models using Python, enhancing production system capabilities.
• Recognized for outstanding contributions to a lung cancer detection and classification project, leading to a promotion.
• Led improvements in work processes and mentored teams on multiple computer vision projects as well as NLP projects, enhancing project outcomes and operational efficiency.
• Developed a CI/CD pipeline, improving development workflows and deployment efficiency.

• Developed algorithms for extracting, creating, and analyzing unstructured data using Python and advanced data engineering techniques.
• Achieved a 98% accuracy in object detection and classification in images using deep learning. This involved designing a custom CNN that utilized transfer learning techniques from models like VGG16, ResNet, and MobileNet, successfully classifying 29 out of 30 images.
• Implemented real-time computer vision pipelines for object detection, tracking, and recognition, enhancing operational response and accuracy.
• Contributed to the NLP team by developing the core CNN model and dataset for a generative chatbot. Employed a Vanilla Transformer architecture and experimented with Byte-Pair Encoding and one-hot encoding techniques for tokenization.