• Enhance geocoding models by fine tuning BERTs model for better accuracy on named entity recognition
• Conduct thorough analysis of underperforming cases and implement feedback loops derived from operational activities to enhance geocoding accuracy
• Implement and deploy computer vision models using Pytorch, Dinov2 as vision transformers and Resnet + Unet as pretrained model. This include classification and segmentation models to estimate vehicle utilization
• Evaluate model performance with hypothesis testing and statistical analysis
Achievements
• Increase computer vision classification accuracy from 83% -> 96%, segmentation dice score 87% -> 95% by preprocessing and fine tune pretrain model
• Increase geocoding accuracy from 96% -> 99.3% by implement feedback loops

I’m a graduate in MSc Business Analytics and Big Data with five years of professional work experience. Passionate about working with large amount of data and to turn those data into information, information into insights and insights into business decisions. I also have a keen interest in technology and how to apply technology to increase business efficiency. Skilled, creative and innovative.
![]() |
![]() |
![]() |
- Age: Young 😁
- Residence: Vietnam
- Address: Ho Chi Minh City
- Phone: +84 984 921 095
- E-mail: tung@vutung.com
- Type: A typical ISTP