Hi, I'm Dhruv Jhang

Manual number plate identification is slow, error-prone, and impractical at scale for traffic and security applications.

Implemented K-Nearest Neighbors (KNN) for character recognition with optimized feature extraction. Processed and annotated over 1,000+ images to train and validate the model for real-world performance.

Reduced misclassification by 15%, demonstrating robust real-world accuracy on diverse plate formats.

Retail investors lack intuitive, data-driven tools for stock analysis and portfolio decision-making.

Designed a stock analysis app prototype for a 24-hour Design Hackathon. Conducted user research and competitor analysis; created wireframes, personas, and user journey maps using Figma to enhance usability.

Secured Runners-Up position among 10+ teams. Delivered a data-driven product case study in under 24 hours by collaborating cross-functionally with designers and mentors.

Raw election data spanning thousands of records lacks accessible, visual insights for voters and researchers.

Collected and analyzed 10,000+ election records using web scraping with BeautifulSoup and data cleaning with Pandas. Built interactive dashboards in Power BI to visualize voter turnout, candidate performance, and party trends.

Delivered actionable insights revealing key correlations between demographic data and voting patterns.

Manual X-ray diagnosis for knee osteoarthritis is subjective, time-consuming, and varies between clinicians.

Built a deep learning pipeline for detecting and classifying knee osteoarthritis severity from X-ray images. Utilized CNNs for automated feature extraction and Random Forest for refining severity predictions.

Achieved 87% model accuracy, demonstrating potential for clinical decision support in orthopedic diagnostics.