A cutting-edge AI-driven system that automatically applies to jobs, tailors resumes, and generates personalized cover letters to optimize ATS compatibility. It scrapes job listings, extracts key requirements, and dynamically adjusts application materials, ensuring a higher match score and improved recruiter response rates. The system also tracks application statuses for continuous improvement.

A scalable cryptocurrency analytics platform built with React, Node.js, and MongoDB, integrating real-time market data, sentiment analysis, and social metrics from sources like CoinMarketCap, Google News, and Reddit. Automated data pipelines with cron jobs ensure continuous updates, high performance, and efficient handling of large-scale real-time data.

A full-stack web application featuring an Express.js backend and an Angular frontend with Bootstrap for a responsive UI. Integrated Ticketmaster APIs for secure data retrieval while ensuring quality compliance. Deployed on Google Cloud Platform (GCP) for scalability and global accessibility.

A classifier to distinguish Martian terrain images with frost, includes training a CNN with MLP, and employing transfer learning with pre-trained models like EfficientNetB0, ResNet50, and VGG16. Evaluation is based on comparing the performance of transfer learning with the CNN + MLP approach.

A multi-label classification on the Anuran Calls (MFCCs) dataset using SVMs with Gaussian and L1-penalized kernels. Optimized hyperparameters via 10-fold cross-validation and evaluated with Hamming score, Exact Match, and AUC. Applied SMOTE for class imbalance and explored Classifier Chains for improved prediction.

Applied SVM, K-means, and spectral clustering on the Breast Cancer dataset to explore supervised, semi-supervised, and unsupervised learning techniques. Evaluated model performance using accuracy, precision, recall, F1-score, and AUC. Compared results across different learning approaches to assess their effectiveness.