Imagination Trumps Knowledge! 

The Naveye project addresses accessibility gaps in current navigation tools, focusing on empowering visually impaired, colorblind, and cognitively challenged users to navigate independently and confidently. By leveraging AI-driven features, it offers innovative solutions such as low-obstacle routes, landmark detection, and a colorblind mode, ensuring inclusivity and usability. Targeting a market of 93 million U.S. adults at risk for serious vision loss, the project emphasizes user-centric design through tailored personas and real-world pain points. Naveye is now a prototype-ready product, showcasing its potential to redefine navigation for underserved communities.

Developed an innovative safety app designed to enhance public safety by enabling real-time sharing of live camera, microphone, and location data with friends or law enforcement, along with SOS alerts. Leveraged wireframing skills to design a user-friendly interface and utilized advanced large language models (LLMs) like BERT and DistilBERT to optimize crime data extraction from police reports. By implementing hyperparameter optimization techniques, improved the model accuracy, increasing the Exact Match score from 59.2% to 63.3% and the F1 score from 0.5885 to 0.6246. This project demonstrates my expertise in product design, AI integration, and applying LLMs to solve real-world problems effectively.

Led the development of an autonomous drone for package delivery, funded by the E-Cell of NIT Hamirpur during my first year of college as part of my first startup. Successfully developed a prototype and engaged in discussions with the government for additional funding to deliver packages in the mountainous regions of Himachal Pradesh. Unfortunately, the startup ceased operations due to COVID-19. Leveraged Robot Operating System (ROS) and Gazebo to simulate drone operations, implementing advanced path planning, image processing, and Proportional Integral Derivative (PID) algorithms to design the Attitude and Position Controller. This experience highlights my entrepreneurial spirit, ability to secure funding and partnerships, and expertise in product design and technical innovation, showcasing strong potential for product-focused roles.

Designed and developed an advanced image captioning system that combines computer vision and NLP, leveraging transfer learning with pre-trained CNN models (VGG-16/Inception) for feature extraction and LSTM networks for generating captions. Built a robust neural architecture comprising an Image Feature Encoder, Sequence Decoder, and Caption Generator, processing over 1,500 images from the Flickr8k dataset using TensorFlow, Keras, and GloVe embeddings. This project highlights my expertise in deep learning, computer vision, and natural language processing, showcasing proficiency in building end-to-end AI-driven solutions.