AI-Powered PDF Processing for UPS & FDX Contracts
I led the development of an AI-powered PDF processing application as a Proof of Concept (POC) for a client, focusing on the intricate task of extracting and processing data from UPS and FDX contracts. This project combined cutting-edge OCR technology, advanced PDF mining techniques, and the latest in LLM prompt engineering to deliver a comprehensive solution.
Key Contributions:
- Advanced OCR & PDF Mining: Leveraged a combination of OCR tools and specialized PDF miners such as Camelot and PuMuPDF to accurately extract data from complex, structured documents.
- LLM Prompt Engineering: Implemented and fine-tuned prompt engineering techniques to enhance the accuracy and relevance of extracted information.
- Streamlit Integration: Developed a cheap and intuitive user interface using Streamlit, enabling seamless interaction with the AI models and easy access to review the extracted data.
- Client Collaboration: Engaged closely with the client to gather detailed requirements, ensuring the solution aligned with their specific business logic and operational needs.
- Comprehensive Reporting: Delivered a 40-page data science report detailing the methodologies, findings, cost analysis, and recommendations, providing the client with deep insights into the processing of their contracts.
Key Contributions:
- Advanced OCR & PDF Mining: Leveraged a combination of OCR tools and specialized PDF miners such as Camelot and PuMuPDF to accurately extract data from complex, structured documents.
- LLM Prompt Engineering: Implemented and fine-tuned prompt engineering techniques to enhance the accuracy and relevance of extracted information.
- Streamlit Integration: Developed a cheap and intuitive user interface using Streamlit, enabling seamless interaction with the AI models and easy access to review the extracted data.
- Client Collaboration: Engaged closely with the client to gather detailed requirements, ensuring the solution aligned with their specific business logic and operational needs.
- Comprehensive Reporting: Delivered a 40-page data science report detailing the methodologies, findings, cost analysis, and recommendations, providing the client with deep insights into the processing of their contracts.
Lvov 
A little busy