Document Automation Platform Development for a Mortgage ISV
The client is a leading mortgage document processing independent software vendor. The ISV’s end-to-end processing system had become increasingly inefficient due to the operational complexity of integrating multiple document processing providers over time. This resulted in multiple disconnected processes, systems and data sources, reduced accuracy rates and increased manual effort to correct the errors. The separation of business lines and dependence on third-party vendors limited the client’s visibility into processes and where to make improvements. When they could identify an issue, long lead times impeded functional enhancements.
Build a centralized document processing platform for multiple systems and modules.
The ML-based platform should automatically recognize, classify, process and index millions of documents and extract information in order to reduce manual correction of classifications and extractions at the backend.
Built an end-to-end document automation platform that…
- Automates intelligent capture, indexing, data extraction, preservation and retention of 700+ mortgage document types
- Ensures faster processing of diverse types of loan packages based on different business SLAs
- Includes dashboards providing key business insights like accuracy rates based on each doc type, field-level extraction, routes to other providers based on business intelligence and customized business rules
- Supports different intake channels (emails, SFTP, APIs)
- Includes data classification, data extraction, data comparison and auditing engines
- Scales dynamically based on the incoming load
- Is capable of retraining ML models based on user and system feedback
- Provides an end-to-end workflow for the backend support team with more than 500+ users active at a time
Developed a robust processing engine by employing Tesseract and Textract OCR solutions to extract information with maximum efficiency.
- Reduced backend support costs by 40% due to higher accuracy rates
- Achieved almost 40% reduction in infrastructure costs with automation
- Reduced mortgage processing time for customers
- Achieved 70% reduction in third-party system licensing costs
- Built centralized document processing platform, reducing inefficiencies due to distributed and outdated systems, manual effort and slow turnaround
- Developed inhouse solution to reduce third-party interference; provide greater control over enhancements; and automate deployment, reducing processing costs and implementation times
- Improved platform monitoring and reporting capabilities with custom dashboards and detailed BI reports in Tableau
AWS | AWS Textract | Tesseract | Machine Learning techniques | MongoDB | Kafka | Java | Python | Angular | Tableau