🚀 Projects

[GitHub Link]

1. 📄 Intelligent Data Extraction from Documents using CV and NLP

• Developed a specialized Document Scanner Web App with a custom Named Entity Recognizer (NER) focused on business card extraction.
• Project phases: setup, data preparation, NER data labeling, preprocessing, model training, and prediction.
• Manually labeled data using BIO tagging for ML model training.
• Built a pipeline for text parsing, visual rendering with Displacy, and bounding box annotations on images.

2. 🛍️ Enhancing Product Categorization with Transformer-Based Attribute Prediction

Part of IndoML 2024 @ BITS Pilani Goa Campus – Ranked 9th out of 400+ teams [link](#)

• Utilized transformer models (BERT, LLaMa-2-7B, GPT-J) for multi-task classification and generative tasks.
• Integrated FAISS-based data augmentation; best performance with GPT-J (6B parameters).
• Handled overfitting with hyperparameter tuning and adaptive learning.
• Applied to real-world multi-label e-commerce classification, emphasizing scalability.

3. 🏥 Sequential Sentence Classification in Medical Abstracts using NLP (Major Project)

• Implemented Bi-ANN, Bi-LSTM, and BERT for sequential classification of medical abstracts.
• Built pipelines for cleaning, tokenizing, and feature extraction from raw text.
• Achieved 92% accuracy via fine-tuning and optimization.

4. 🤖 Colour Sorting of Objects with 3D Robotic Arm using OpenCV (AMURoboclub)

• Designed a 3D robotic arm to sort objects by color.
• Applied OpenCV for robust hue-based object classification.
• Used Raspberry Pi to control robotic movements precisely.

5. 🏠 House Price Prediction Web App using Machine Learning

• Developed an intuitive web app to predict housing prices.
• Trained a regression model and deployed it via a pickle file for fast inference.
• Designed front-end using HTML/CSS for a smooth UX.