Projects

Built a time-sensitive itinerary summarizer using Named Entity Recognition, POS tag, SpaCy, nltk, and dependency parser. The proposed method had a 20% increased BLEU score compared to the baseline approach. Colab Presentation

Analyzed vehicular trajectory data in Python to identify the scope of errors between real-time data and simulated data. Prepared graphs and visualizations after data processing. Results showed a significant difference in acceleration prediction based on RMSE. Code  

Designed mobile application ShakeIt to increase awareness about earthquake preparedness before, during and after the disaster. Proposal Presentation 

Built a sliding puzzle game 2048 Solver. Code  Presentation 

Built a Figma Prototype in collaboration with LexisNexis to aid jurors regarding virtual evidence viewing. Processed data and published results about participants doing consistently well with the prototype in CSCW 2022. Paper Presentation

Designed software to persuade users to be more eco-friendly in their regular activities. Managed data collection process and performed thematic analysis to report how participants preferred the prototype in the form of a research paper. Figma Prototype

Built a Arduino-based system for automatic water tank management using solar panels. Demo 

Implemented DQN, ES, and PPO (Deep Reinforcement Learning techniques) to ensure an agent turns left in an intersection without collisions with through-traffic. Results showed PPO had the best average evaluation rewards. Colab

Collaborated on implementing Vision Transformer model for crop classification on spatio-temporal data with 95% accuracy. Presentation

Proposed health transaction analytics report inclusion to existing financial wellbeing application. [Slides cannot be shared for privacy reasons]