Adityasinh

Created a full-stack stock-trading web application where users can practice trading with a $10,000 virtual balance. It integrates the Yahoo Finance API for company summaries and price trend visualizations, with a Django backend managing login, registration, stock searches, and transactions. Python was used to configure the backend, PostgreSQL database was used to store user information and trade details, and the frontend was created using React.js.

My Machine Learning CUDA project focused on leveraging NVIDIA's CUDA platform to accelerate machine learning algorithms using GPUs. I developed highly optimized versions of common machine learning tasks, specifically a k-Nearest Neighbors (k-NN) algorithm and a linear regression model. The k-NN implementation was designed for parallel processing, allowing for very fast and accurate queries. For linear regression, I incorporated a per-feature gradient reduction technique to achieve efficient performance. A key part of the project involved building a complete pipeline on an NVIDIA Jetson Orin, handling data from initial parsing to managing memory between the CPU and GPU.

Developed an interpreter for SFort95 by creating lexical analysis and parsing algorithms to accurately translate code into executable instructions. Implemented robust error handling to detect and report syntax errors, improving reliability. Optimized the interpreter's performance through algorithmic enhancements, resulting in faster execution and increased efficiency.

Developed a Minesweeper game with an AI agent capable of efficiently solving the board. The game includes a class that represents the Minesweeper board, initializing it with a specified height, width, and number of mines. Mines are placed randomly, and the board's state is represented textually. The AI agent employs logical reasoning to deduce safe moves and identify mine locations. It maintains knowledge of the game state through sentences that represent logical statements about the board. By leveraging this knowledge, the AI marks cells as safe or mines and updates its knowledge base, ensuring a strategic approach to solving the game. The project demonstrates a robust implementation of game mechanics and AI problem-solving techniques.