My Projects

ChatNexus: Large Language Model using PyTorch

• Developed a GPT-based Language Model using PyTorch, leveraging multi-head self-attention mechanisms and transformer blocks to enhance text generation capabilities, optimizing model performance on both CPU and GPU.
• Implemented a data processing pipeline using Python, and Pickle for efficient handling of large text datasets, incorporating techniques like token embedding and positional encoding.

VectorDriveAI: PyTorch and PostgreSQL

• Developed a secure data classification system using GPT-based models and PGVectorStore, integrating Google Drive with options for OpenAI API or Ollama for sensitive data detection and classification.
• Implemented a scalable vector search solution with PostgreSQL pgvector, supporting both OpenAI and Ollama embeddings, for efficient document retrieval and real-time sensitivity scoring.

Super Health App: Received 1st price for UIC SparkHacks

• Developed Super Health App, a multifunctional health management platform, integrating telemedicine consultations, prescription management, reminders, and a mindful space, using Python and Tkinter.
• Overcame challenges such as SQL database integration, user input handling, and game integration, while achieving a working prototype within the SparkHacks-2024 hackathon timeframe, showcasing adaptability and technical proficiency.

Bertram Capital Project

• Engineered a Coffee Expenditure Management Application Prototype for Bertram Capital employees using Python and Tkinter, facilitating individual tracking of coffee preferences and expenditures with SQLite3 database integration.
• Implemented intuitive GUI features for order processing and financial overview, including automated debt calculation and equitable bill assignment based on historical payment data.

fetch_test

• Developed an Android application using Retrofit and Gson for efficient API interactions, employing a singleton pattern for the API client to ensure optimal resource use and consistent data access.
• Implemented a RecyclerView Adapter to display data from a RESTful API in an Android app, optimizing list performance and memory management through Android's ViewHolder pattern.

Which Way: Game

• Developed 'Which Way?', a graphical Python game based on snakes and ladders for CS 111, implementing interactive gameplay and object interactions using the Turtle module.
• Designed user interfaces and game mechanics, including ladders, snakes, and victory conditions using Python, enhancing user engagement through intuitive game flow and state management.

Contact Me

ayush975600@gmail.com

(224) 358 8722

Download CV