Haseeb Ul Hassan

Develop custom AI and ML solutions for enterprise clients, including data augmentation and model fine-tuning to improve model performance and delivery speed.
Built ForecastIQ, an end-to-end demand-forecasting and planning pipeline that centralized inputs, automated daily processing, and eliminated manual spreadsheet work, saving $100K+/year in labor and reducing excess inventory.
Lead ongoing ML initiatives across PromoIQ, automation, and deployment workstreams to improve operational efficiency and speed up delivery of client-facing solutions.
Led and mentored a team of ML interns, ran daily standups, assigned work, and conducted weekly demos to improve execution speed and build team capability.
Delivered custom AI solutions for Kofile, automating image-based cataloging and data extraction across multiple categories to reduce manual entry, improve throughput, and increase data accuracy.
Manage and update APIs for ongoing AI services, ensuring seamless integration and functionality.
Implemented generative AI workflows for client and internal use cases, enabling automation and expanding product capabilities across internal tools and client deliverables.

Develop custom AI and ML solutions for enterprise clients, including data augmentation and model fine-tuning., Built ForecastIQ, an end-to-end demand-forecasting and planning pipeline that saved $100K+/year in labor and reduced excess inventory., Lead ongoing ML initiatives across PromoIQ, automation, and deployment workstreams., Led and mentored a team of ML interns, ran daily standups, assigned work, and conducted weekly demos., Delivered custom AI solutions for Kofile, automating image-based cataloging and data extraction., Manage and update APIs for ongoing AI services., Implemented generative AI workflows for client and internal use cases.

Trained PaddleOCR Detection model by converting custom dataset labels, enhancing text detection performance.
Fine-tuned LayoutLMv3 for sequence classification, tuning batch size and learning rate to improve model accuracy on document understanding tasks.
Trained and compared detection and segmentation models, converted segmentation labels to detection format, and applied augmentation techniques to improve accuracy and generalization.
Developed and tested Android-based mobile applications integrating computer vision models for detection and segmentation workflows.

Trained PaddleOCR Detection model by converting custom dataset labels., Fine-tuned LayoutLMv3 for sequence classification., Trained and compared detection and segmentation models., Developed and tested Android-based mobile applications integrating computer vision models.