Hi, I am Prajwal Negi

This project leverages the LangChain framework integrated with OpenAI's language model to build an intelligent restaurant name and menu generator. The system accepts a cuisine type as input and then generates a creative and fancy restaurant name tailored to that cuisine. Following that, it automatically suggests a list of relevant menu items, suitable for the generated restaurant name.

Chat App for the user where user can create a private room and can chat with other friends with same roomId. Websocket is use for providing a full duplex communication between users.

This is a project where user can get his live location, updated every few seconds. Leaflet is use for accessing the street map.

This project is a real-time object detection system built using YOLOv8, integrated with OpenCV for video processing and SMTP email alerts for notifying unusual object presence. The application uses a webcam feed to detect objects frame-by-frame with the help of a pre-trained YOLOv8 model and overlays bounding boxes along with FPS metrics on the live video. When objects are detected, an alert email is automatically sent to a configured recipient, ensuring prompt security notification.

This project is a smart fitness monitoring system built using Ultralytics' AI Gym solution, leveraging YOLOv8 pose estimation to analyze human posture and movements during workout routines like push-ups and pull-ups. By processing pre-recorded videos, the system accurately tracks key body landmarks (specifically joints like shoulders, elbows, and wrists) and monitors repetitions. The system visually overlays pose estimation outputs on each frame, counts repetitions, and generates annotated output videos for review and feedback.

This interactive virtual coffee machine utilizes hand tracking to allow users to make coffee selections using finger gestures without any physical touch. Using a webcam feed, the system detects specific finger patterns to navigate through multiple selection stages such as choosing the type of coffee, sugar quantity, and milk preference. The gesture-based interaction is visualized with an engaging UI, showing animated option selection using a progress ellipse. It provides a touchless and futuristic user experience, ideal for smart cafés.

This real-time computer vision project enables users to control system volume using simple hand gestures—specifically the distance between the thumb and index finger. By tracking hand landmarks with a webcam, it dynamically adjusts the volume based on finger distance, displaying a visual bar and percentage overlay.

A real-time quiz system where users can select answers using hand gestures instead of a mouse or keyboard. The system uses a webcam to detect hand landmarks and allows interaction with multiple-choice questions by tracking finger gestures. By pinching the index and middle fingers together over a selected answer, users can register their choice. A visual progress bar shows quiz completion, and a final score is displayed upon completion.

This project implements a virtual keyboard that allows users to type by interacting with an on-screen QWERTY keyboard using hand gestures captured from a webcam. Using computer vision and hand landmark detection, it tracks finger positions to detect which key the user is 'hovering' over and simulates key presses when a pinch gesture (distance between the index and middle finger) is detected.

This is a console-based application developed in C++ to manage hostel bed reservations efficiently. The system allows students to reserve beds by entering their details, which are then stored in a file for record-keeping. It tracks available beds, updates hostel data, and ensures reservation logic through file handling operations. The project demonstrates practical knowledge of object-oriented programming, file I/O in C++. It’s a simple yet effective solution for managing hostel accommodations in small institutions.

This project predicts Tesla's stock trading volume using historical stock data, including prices and dates. By analyzing the data with visualizations, it explores trends and relationships between stock features and volume. Two machine learning models—Linear Regression and Random Forest Regressor—are used.

A Convolutional Neural Network (CNN) was trained for plant disease classification using a dataset of images categorized into three classes: Healthy, Powdery, and Rust. The model architecture included three convolutional layers followed by max-pooling, flattening, and several dense layers. The model is trained using the Adam optimizer and categorical cross-entropy loss. During training, the accuracy steadily improved from 39.78% to 97.84%, with a final validation accuracy of 90%. Evaluation on the test set yielded a test accuracy of 93.69% and a loss of 0.2332.

This project implements a deep learning model using LSTM (Long Short-Term Memory) networks to predict the next word in a given text sequence. A small corpus of motivational and philosophical phrases is used as training data. The text is tokenized using Keras's Tokenizer, and input sequences are generated in an n-gram style. The model architecture includes an Embedding layer to transform word indices into dense vectors, followed by an LSTM layer with 150 units to capture sequential dependencies. A final Dense layer with softmax activation is used to output the probability distribution over the vocabulary.After training the model for 100 epochs using categorical cross-entropy loss and the Adam optimizer, it can predict the next likely word for any given input phrase.

This project is a ML-based system that predicts whether a customer will make a purchase based on their demographic and behavioral data. It utilizes Logistic Regression, a supervised learning algorithm, to classify customer intent based on features such as age, time spent on the website, whether a product was added to the cart, gender, and past purchase behavior. The program accepts real-time user input to simulate customer behavior and predict their likelihood of making a purchase.

This project is a responsive EdTech Dashboard designed to provide users with an overview of their learning progress through visually appealing and interactive charts. The dashboard features user profile information, badges, past courses, projects, and recommended courses — all integrated into a clean, intuitive interface.The core highlight is the use of Chart.js to render dynamic visualizations:Line Chart: Displays hours spent learning per day over several months.Bar Chart: Illustrates the number of questions solved per month.The dashboard incorporates a responsive navigation menu with icons (via Font Awesome) and uses semantic HTML for accessibility. The design supports smooth user interaction and visual tracking of performance metrics for learners.