🧠 My Blog

One of my earliest and most exciting projects was building a 15-Puzzle Solver in Python — a classic sliding puzzle where tiles numbered 1 to 15 must be arranged in order. My goal was to create an intelligent solver that could handle even the most complex puzzle states efficiently.

I implemented powerful search algorithms like A* (A-star) and Breadth-First Search (BFS) to explore the puzzle state space and find the shortest path to the solution. I also integrated heuristics such as Manhattan Distance to guide the solver intelligently, significantly improving its speed and efficiency. Through this project, I learned how AI search algorithms work, how heuristics can improve decision-making, and how computers "think" through problems step by step.

After building the logic behind the 15 Puzzle Solver, I wanted to make it more interactive and fun. That's why I created the 15 Puzzle GUI Solver, a graphical version of the solver using Pygame.

With this version, users can visually see the puzzle state, interact with it, shuffle tiles, and even watch the AI solve it in real-time. This project taught me how to connect complex backend algorithms with user-friendly interfaces — a crucial skill for building AI-powered apps. It also gave me hands-on experience in event handling, animations, and UI design.

Managing schoolwork can be challenging, so I built the Homework Tracker — a productivity tool designed especially for students like me. It helps organize homework, projects, and tasks by tracking subjects, descriptions, due dates, priorities, and log dates.

I built it using Python and CSV for data handling, and added a colorful GUI with Pygame to make it visually engaging. To make it even smarter, I integrated features like email and Telegram reminders, so you never miss a deadline. This project helped me explore real-world software design — combining data storage, user interaction, and automation in one practical app.

The Dream Insights project explores a fascinating field: analyzing dream descriptions using machine learning. The goal was to build a system that predicts factors like lucidity, emotional intensity, and realism based solely on text.

I trained machine learning models like Random Forest Regression and used Principal Component Analysis (PCA) for visualizing high-dimensional data. I also created a web interface so users could input dream descriptions and instantly receive predictions. This project taught me how to handle natural language data, preprocess text, train regression models, and build interactive AI tools. It's one of my most creative and exploratory projects so far.

The Scam Detector is a powerful project focused on identifying scams within video transcripts using Natural Language Processing (NLP). I built a pipeline that analyzes text data, performs vectorization, and detects scam-related keywords and patterns.

Using ML models and carefully designed scam indicators, the system flags potentially fraudulent or harmful content before it spreads. This project deepened my understanding of NLP, feature engineering, and text classification. It also showed me how AI can be applied to real-world safety problems — protecting people from online scams and misinformation.

Plant Doctor AI is one of my most impactful projects — a deep learning-powered application that identifies plant diseases from leaf images. Using a ResNet18 CNN model trained on the PlantVillage and PlantDoc datasets, the system provides accurate disease predictions to help farmers, researchers, and plant enthusiasts.

The project features a beautiful web interface, camera support, Grad-CAM heatmap visualization for explainable AI, and deployment on Render with a custom domain. This project taught me how to train deep learning models, integrate them into a full-stack application, and deploy AI systems for real-world use.