Dharak P- Senior Data Scientist

Major Projects

AI-Powered Car Damage Cost Estimation – Visual Damage Analysis and Cost Prediction

Developed a computer vision system to assess car damage and estimate repair costs from multi-angle images. Built logic to convert normalized coordinates to original image space for accurate damage localization. Implemented intersection analysis algorithms to determine severity by overlapping damaged areas with car parts. Automated CSV report generation with detailed cost estimations, enabling faster insurance claim processing and accurate resale evaluations for second-hand vehicles.

Germany

Backend:

Python

Computer Vision:

OpenCV
Shapley
YOLO

Financial Guide – Personal Finance Insights and Goal-Based Saving Recommendations

Created a finance management tool that analyzes bank statements to provide actionable spending insights. Built algorithms for automated expense categorization and personalized saving suggestions based on user habits. Designed goal-based strategies to help users plan for major purchases like a home or car by calculating monthly savings targets. Enabled visual financial reporting to enhance clarity and support better money management decisions.

USA

Backend:

Python

AI/ML/NLP:

LLM integration
PDF Parser
Plotly

Vessels Predictive Maintenance – AI-Driven RUL Estimation and Maintenance Optimization

Built an AI-powered system to predict vessel maintenance needs using real-time sensor data. Developed robust logic to calculate critical components' Remaining Useful Life (RUL) and generate timely repair or replacement recommendations. Led end-to-end data handling, from acquisition to preprocessing, to ensure accurate forecasting. Enabled proactive maintenance planning through a real-time monitoring interface, reducing unplanned downtime and extending vessel lifespan.

Australia

Backend:

Python

AI/ML:

Clustering algorithms

Property Recommendation System – Personalized Real Estate Search Using Multi-Parameter Matching

Engineered a smart recommendation system that analyzes user-defined inputs like location, budget, and preferences to identify suitable real estate listings. Designed an intuitive input interface and implemented algorithms to process multi-parameter queries efficiently. Built logic to evaluate property data and deliver real-time, personalized recommendations, streamlining the property search process and enhancing user satisfaction through tailored suggestions.

UK