Life Lines of Celebrities_GENAI

Explore the biographies, milestones, and philosophies behind influential figures. Learn from their challenges and philosophies to fuel your own ambitions and personal growth.

• Code

NutriScan: Your Smart Food Analyzer_GENAI

NutriScan is your pocket nutritionist. Snap a photo of your food for instant nutritional details – calories, macros, and more. Understand your diet, make healthier choices, and track your intake with ease. Perfect for college students and mindful eaters, deployed on GCP.

• Project Live
• Code

Jay Fast Foods - NutriNet ChatBot_NLP

I developed an end-to-end NLP chatbot specifically designed to take food orders. This chatbot leverages Dialogflow for natural language understanding, allowing users to easily place orders for items like Pizza, Chole Bhature, Masala Dosa, and more. I used FastAPI to build the backend, ensuring a fast and efficient API, and MySQL for storing important order details.

• Code
• Video Explanation

JaySummBot- Text Summarizer_NLP

I developed an end-to-end text summarization project using the Google Pegasus model. For deployment, I used Docker, AWS EC2, and ECR, and gained experience with CICD practices. To share my knowledge, I created a YouTube video explaining the project's components and their roles.

• Code
• Video Explanation

Kidney Disease Classification (Tumor or Normal)_DL

I built a kidney disease classification system to detect tumors, using the VGG16 model, modular Python code, and MLflow for experiment tracking. Inspired by Krish Naik's tutorials, I also used DVC for streamlined project workflow management and deployed in AWS.

• Code
• Video Explanation

Quora-Duplicate-Questions-Project_NLP

I competed in the Quora Duplicate Questions Kaggle challenge, classifying question pairs. Through feature engineering and algorithms like RandomForest and XGBoost, I progressively improved model accuracy from 68% to ~80%.

• Code

Movie-Recommendation-System_using_NLP

I developed a movie recommendation system using NLP techniques. I preprocessed text data (genres, keywords, cast, crew), including stemming and the creation of a 'tags' column. Using 'Bag of Words' and cosine similarity, I built a model that recommends movies based on their content similarity

• Code

Dog-Vision Project_DL

I built a dog breed classifier using the MobileNetv2 pretrained deep learning model. I preprocessed images using functions like os.listdir, to_numpy(), and numpy.argmax. With a dataset of 10k images, I trained on 3k images and achieved an impressive 80% accuracy.

• Code