Objective: to provide users with creative, AI-generated suggestions for converting various waste items into valuable resources, promoting sustainable waste management. 

Technologies Used: LLaMA (Open LLM), Hugging Face Transformers, PEFT, BitsAndBytes, Gradio, Google Colab 

User Input for Waste Material: Users can input a waste item (e.g., plastic bottles, old electronics) to receive tailored repurposing ideas. 

Fine-Tuned Model for Idea Generation: Utilized LLaMA-2-7B, fine-tuned to generate context-specific ideas for repurposing waste items into useful products or resources.

Interactive UI: Built with Gradio, providing a user-friendly interface where users can enter waste items and view AI-generated ideas in real-time.

Efficient Model Optimization: Incorporated 4-bit quantization with BitsAndBytes for resource-efficient deployment on limited hardware.

Data Preparation: Collected and organized waste material data into a structured CSV file for model training.

Model Training: Fine-tuned LLaMA-2-7B on Google Colab using PEFT and customized hyperparameters to improve response relevance and creativity.

UI Integration: Developed an interactive Gradio UI, allowing users to input waste materials and receive instant AI-driven ideas.

Future Enhancements: Planning to integrate an image generation model to create visuals based on AI-generated repurposing ideas.

Objective: To allow users to input their preferences and receive tailored news articles and summaries based on their interests. 

Technologies Used: Streamlit, Googlesearch-Python, BeautifulSoup, gTTS, Groq API 

User Input for Preferences: Users can specify their country, news category (e.g., technology, business), specific news sources (e.g., BBC News), and professional roles (e.g., student, lawyer) to generate personalized news queries.

Google Search Integration: Utilized Googlesearch-Python to search for relevant news articles based on user queries and retrieve the top articles matching their preferences.

Article Summarization: Users can select articles, and the app scrapes content using BeautifulSoup to extract and summarize the main points (first three paragraphs).

Text-to-Speech Functionality: The app converts summaries into audio files using gTTS (Google Text-to-Speech), allowing users to listen to summaries directly within the app.

API Integration: Integrated Groq API, serving as a placeholder for potential future enhancements with AI summarization tools.

User Preferences: Users input preferences through a Streamlit form (country, category, source, role).

News Search: The app generates search queries based on user inputs and retrieves relevant news articles using Google Search.

Article Selection: Users select articles, and the app scrapes and summarizes the content.

Audio Summary: The app employs gTTS to generate audio versions of summaries for user convenience.

Objective: To accurately classify the health of rice leaves and detect diseases using advanced image processing and machine learning techniques.

Technologies Used: Digital Image Processing, Machine Learning Algorithms, Python

Image Processing: Applied various digital image processing techniques to preprocess and enhance images of rice leaves.

Machine Learning: Experimented with the Support Vector Machine (SVM) algorithm to improve model performance for disease detection and classification.

Graphical User Interface: Built a user-friendly GUI using Python libraries for easy interaction with the model.

Image Processing Techniques: Utilized techniques such as filtering, segmentation, and feature extraction to prepare images for analysis.

Support Vector Machine: Implemented and fine-tuned the SVM algorithm to achieve higher accuracy in classification and disease detection.

Python Libraries: Leveraged Python libraries like OpenCV for image processing, scikit-learn for machine learning, and Tkinter for GUI development.

Objective: To create a user-friendly platform for hostel residents to submit and track complaints, and for administrators to manage and resolve them efficiently.

Technologies Used: HTML, CSS for front-end development, and PHP for back-end development.

User-Friendly Interface: The web application features a clean and intuitive design, making it easy for residents to navigate and submit complaints.

Efficient Complaint Management: Administrators can quickly access, track, and manage complaints, ensuring timely resolutions.

Front-End Development: Utilized HTML and CSS to create a responsive and visually appealing user interface.

Back-End Development: Implemented server-side logic using PHP to handle complaint submissions, database interactions, and administrative functions.