Vansh Sharma

I’m an aspiring Data Scientist and analytics professional specializing in AI/ML, with hands-on experience turning data into decision-ready insights. I focus on data visualization, exploratory data analysis (EDA), dashboard development, and reporting automation, always with the goal of solving real-world business problems through data-driven thinking.

In my Data Science Intern role at AM Consulting Co. WLL, Bahrain (Remote), I developed interactive Power BI dashboards with KPIs, filters, slicers, drill-downs, and performance visualizations to support smarter decisions. I handled data preprocessing, cleaning, transformation, and validation using Python, Pandas, and NumPy, and I also built automation workflows in Arbutus Analyzer to improve reporting efficiency. To keep reporting consistent and streamlined, I created automated multi-sheet Excel reports using OpenPyXL and supported data auditing, exception analysis, and business rule validation to maintain data quality standards.

As a Data Science Intern at Substratal Solutions Pvt. Ltd. (SSPL), Bahrain (Remote), I worked on Power BI dashboards for business and sales analytics, using interactive elements like filters, slicers, drill-down functionality, and KPI-based reporting. I created relationships across multiple data sources, implemented DAX calculations for dynamic reporting, and contributed to iDashboards solutions for business performance monitoring. I also collaborated with senior analysts to strengthen dashboard design principles and enterprise reporting workflows.

In my ML & Research Intern experience at Centre for Health Innovations (MRIIRS), I built an NLP-based depression detection system and achieved 90.04% classification accuracy using the DIAC-WOZ dataset. I implemented and compared USE and ELMo embeddings, then built end-to-end NLP pipelines across preprocessing, embedding generation, model training, and evaluation using accuracy, precision, recall, and F1-score. I’ve also applied these end-to-end ML skills in projects like my diabetes detection system using Logistic Regression, KNN, and SVM, plus EDA with histograms, heatmaps, and pairplots to uncover patterns.