PDF Intake (PDFI) for myUSCIS
Description
Scanned PDFs submitted through MyUSCIS must be validated against form-specific business rules related to both the overall document and the contents of specific fields. Constructed a service that can process a scanned input document and return all information pertinent to these validation rules in a consistent structure (JSON) to a user-facing ELIS microservice. The GenAI powered library utilizes Amazon Bedrock – Anthropic Claude 3.7 Sonnet V1 Foundation Model to extract data from PDF forms. The service provides ability to submit forms online through MyUSCIS UI to Lockbox instead of via mail.
Detailed example
The output of this AI system is structured information about the validation rules applied to the input form as well as the extracted contents of filled fields on the form, presented in a JSON format readable by both humans and machines that is consistent with existing ELIS databases.
AI / analytics pattern
Generative AI: AI that generates new or synthetic content (e.g., images, videos, audio, text, code).
Automation level / stage
c) Deployed – The use case is being actively authorized or utilized to support the functions or mission of an agency.
Expected benefit
Develop a service that can extract relevant fields from a scanned PDF submitted through MyUSCIS and build a JSON as an output to the ELIS microservice. The new service will utilize AWS Bedrock provided foundation model. It is an engineering solution that minimizes development time to add new forms or form revisions with high accuracy.
Controls / human review
ATO: Yes; PIA: https://www.dhs.gov/publication/dhsuscispia-056-uscis-electronic-immigration-system-uscis-elis
Data needed
During development, this system is evaluated using both manually created synthetic data (i.e. filled PDF forms with annotated contents) and production data (scans of forms submitted previously through Lockbox as scanned TIF files). The underlying pretrained foundation model supplied by Bedrock service is used as-is with no further training or fine-tuning.