AI Chatbot Development Using LangChain and Ollama

Developed an AI chatbot leveraging LangChain and Ollama's LLM to provide real-time, context-aware conversational responses. The chatbot features an interactive front-end built with Streamlit, delivering a seamless and engaging user experience. Overcame technical challenges in API integration, input validation, and backend functionality to ensure reliable and efficient performance. This project demonstrates proficiency in conversational AI, front-end development, and effective problem-solving to create intelligent, user-centric applications.

Real-Time Vehicle Detection using OpenCV and YOLOv5

Designed and implemented a high-performance vehicle detection system using the YOLOv5 deep learning model, pre-trained on a dataset of 2,704 traffic scenario images. The system achieved a remarkable 98% detection accuracy by integrating OpenCV for real-time image preprocessing and vehicle recognition. Leveraging transfer learning, the model was fine-tuned to enhance training efficiency and detection precision, making it highly optimized for real-world traffic environments. This project showcases expertise in computer vision, deep learning, and real-time application development for intelligent transportation systems.

Skimlit-project

Engineered an advanced NLP model to classify abstract sentences into their respective roles, such as objectives, methods, and results, facilitating researchers in efficiently skimming through literature and focusing on relevant sections. Experimented with five diverse modeling approaches, including a 1D Convolutional model, a Naive Bayes model with pretrained embeddings, a convolutional model with character embeddings, and a hybrid embedding layer combining pretrained token and character embeddings.

PowerBI Dashboard analysing Road Accidents

A Dashboard is prepared using PowerBI based on road accidents data in India. This analysis includes analysing the major causes of road accidents, which includes Gender of driver, Accident Severity, Count of Casualities based on their age etc..,

Predicting Student Performance Index

A Multiple Regression model was developed to predict the Student Performance Index using key factors such as Hours Studied, Previous Scores, Sleep Hours, and Number of Practice Papers. By employing Multiple Variable Linear Regression techniques, the model effectively learned the relationships between these features and the performance metric. The model achieved impressive results with training, cross-validation, and test costs of 220.20, 230.12, and 226.96 respectively, significantly outperforming the benchmark of 368.51.

Online Food Orders Analysis

Analysis on the Online Food Orders to uncover insights into customer behavior and preferences. Utilized Python frameworks such as NumPy, Pandas, Matplotlib with Seaborn to plot various graphs obtaining relations among data. Analyzed customer feedback to understand their experience with online food services and identified areas for Improvement.