Projects

AI-Powered PacMan Commentary, AIEA Lab, UCSC

Jan 2024 - Present | California, USA

• Engineered a novel approach to analyze Pac-Man gameplay using computer vision to generate spatial representation of game.
• Leveraged advanced LLMs and prompt engineering to generate informative captions for every 5 seconds of gameplay.
• Fine-tuned the model to produce high-quality, contextually relevant commentary, enhancing the overall viewer experience.
  

Here is the - Github Repo

LLM Smart Chat

Nov 2024 | California , USA

• Leveraged a Large Language Model (LLM) (OpenAI) to answer user queries based on the contents of a pdf file.
• Utilized efficient document chunking and retrieval techniques to ensure that only the most relevant sections of the document are used to generate context-specific answers.

Here is the - Github Repo

Road To Chciken Dinner

Nov 2023 - Dec 2023 | California , USA

• Built Multiple Linear Regression model, performed Model selection using AIC & BIC criterion to predict Winning percentile.
• Conducted pair t-tests to assess group significance and utilized p-values to determine the significance of predictors.

Here is the - Github Repo

Image Classification Using Transfer Learning

Nov 2023 - Dec 2023 | California, USA

• Trained a Swin Large Vision Transformers on a custom dataset to classify images into 100 classes.
• Utilized Hugging Face’s PyTorch image models library to build and train the transformer and achieved an accuracy of 82%.
  

Here is the - Github Repo

Eye for the Blind

Jan 2021 - March 2021 | Bangalore, India

• Designed Encoder-Decoder neural network using self-attention and GRU to generate descriptive image captions, empowering visually impaired individuals to comprehend images; achieved accessibility and inclusiveness for 1000+ users.
• Deployed the model on AWS using FLASK with a TAT of 2 seconds.
  

Here is the - Github Repo