Add Strata DocumentAI API app

[doshitan]

Ticket

Resolves #253

Changes

TODO extract useful notes for configuring DocumentAI API

TODO add domain/HTTPS config

TODO update or remove E2E tests

TODO infra test prefix does still cause issues with app-documentai name, PR envs fine (into the 1000s at least, and dependent on env name). We could use app-docuai, which fits the other (current) longest app names.

Context for reviewers

Have not (yet) copied overall the custom templates from #274. A bit unclear if those should be something we just ship as a part of the DDE module in the template vs DocumentAI API specific.

Review all the TODO(pre-merge): comments.

Testing

Video demo of PR environment (public link): https://drive.google.com/file/d/1t3-cBPaE-4i2_XkEPgbSpbBqhIU36elO/view?usp=drive_link

Using https://www.paystubhero.com/wp-content/uploads/2023/11/896-1.jpg

See results:

{
  "jobId": "ab2e401b-e2f1-4e79-8b60-a21562aa07e5",
  "jobStatus": "completed",
  "message": "Document processed successfully",
  "createdAt": "2026-05-07T21:03:47.833187Z",
  "completedAt": "2026-05-07T21:05:03Z",
  "totalProcessingTimeSeconds": 75.17,
  "matchedDocumentClass": "W2",
  "fields": {
    "employerInfo.employerAddress": {
      "confidence": 0.84,
      "value": "41980 Ann Arbor Rd. E Plymouth, NC"
    },
    "employerInfo.controlNumber": {
      "confidence": 0.92,
      "value": ""
    },
    "employerInfo.employerName": {
      "confidence": 0.98,
      "value": "Paystub Hero"
    },
    "employerInfo.ein": {
      "confidence": 0.97,
      "value": "39-3598535"
    },
    "employerInfo.employerZipCode": {
      "confidence": 0.97,
      "value": 48170
    },
    "filingInfo.ombNumber": {
      "confidence": 0.97,
      "value": "1545-0029"
    },
    "filingInfo.verificationCode": {
      "confidence": 0.96,
      "value": ""
    },
    "other": {
      "confidence": 0.95,
      "value": ""
    },
    "federalTaxInfo.federalIncomeTax": {
      "confidence": 0.98,
      "value": 9467
    },
    "federalTaxInfo.allocatedTips": {
      "confidence": 0.96,
      "value": 0
    },
    "federalTaxInfo.socialSecurityTax": {
      "confidence": 0.97,
      "value": 4960
    },
    "federalTaxInfo.medicareTax": {
      "confidence": 0.97,
      "value": 1160
    },
    "employeeGeneralInfo.employeeNameSuffix": {
      "confidence": 0.96,
      "value": ""
    },
    "employeeGeneralInfo.employeeAddress": {
      "confidence": 0.13,
      "value": "41980 Ann Arbor Rd. E Plymouth, CA"
    },
    "employeeGeneralInfo.employeeLastName": {
      "confidence": 0.96,
      "value": "Jesan"
    },
    "employeeGeneralInfo.employeeZipCode": {
      "confidence": 0.98,
      "value": 48170
    },
    "employeeGeneralInfo.firstName": {
      "confidence": 0.97,
      "value": "Abdur Rahaman"
    },
    "employeeGeneralInfo.ssn": {
      "confidence": 0.95,
      "value": "498-74-9874"
    },
    "federalWageInfo.socialSecurityTips": {
      "confidence": 0.96,
      "value": 0
    },
    "federalWageInfo.wagesTipsOtherCompensation": {
      "confidence": 0.96,
      "value": 80000
    },
    "federalWageInfo.medicareWagesTips": {
      "confidence": 0.96,
      "value": 80000
    },
    "federalWageInfo.socialSecurityWages": {
      "confidence": 0.96,
      "value": 80000
    },
    "nonqualifiedPlansIncom": {
      "confidence": 0.98,
      "value": 0
    }
  },
  "error": null,
  "additionalInfo": null
}

Preview environment for app

• Service endpoint: https://p-278-app-dev-746181005.us-east-1.elb.amazonaws.com
• Deployed commit: b30e9f7

Preview environment for app-catala

• Service endpoint: http://p-278-app-catala-dev-861146667.us-east-1.elb.amazonaws.com
• Deployed commit: b30e9f7

Preview environment for app-rails

• Service endpoint: https://p-278-app-rails-dev-980702910.us-east-1.elb.amazonaws.com
• Deployed commit: b30e9f7

Preview environment for app-flask

• Service endpoint: https://p-278-app-flask-dev-918012928.us-east-1.elb.amazonaws.com
• Deployed commit: b30e9f7

Preview environment for app-nextjs

• Service endpoint: https://p-278-app-nextjs-dev-1225001131.us-east-1.elb.amazonaws.com
• Deployed commit: b30e9f7

Preview environment for app-documentai

• Service endpoint: http://p-278-app-documentai-dev-2070334721.us-east-1.elb.amazonaws.com
• Deployed commit: b30e9f7

[doshitan]

@laurencegoolsby could you review and respond to any TODO(pre-merge): comments (well, the ones that are questions, haha)? I came across some things I didn't understand in the course of going through the #274 changes.

[laurencegoolsby]

Reviewed/responded to all TODO(pre-merge): comments.
Created navapbc/strata-template-documentai-api#52 , navapbc/strata-template-documentai-api#53

[laurencegoolsby]

The document block is needed for PDF/document extraction. The image block handles image files (photos of documents). We need both - document for PDFs/TIFFs, image for JPEGs/PNGs.

The document config extracts page-level text with bounding boxes, which is what BDA needs for field extraction.

[doshitan]

Are these document settings highly specific to the Strata DocumentAI implementation? Or are they reasonable general baselines we should just ship as a part of the Document Data Extraction config (like we are doing for the image part)?

[laurencegoolsby]

@doshitan - Reasonable general baselines. Ship as part of DDE.

[laurencegoolsby]

Agree — the app code should use DDE_INPUT_LOCATION etc. instead of DOCUMENTAI_* prefixed vars.

Created navapbc/strata-template-documentai-api#52

[laurencegoolsby]

@doshitan I retract my previous statement. DOCUMENTAI_INPUT_LOCATION makes sense here. The DOCUMENTAI_* prefix makes sense for the applications environment variables.

The DDE module provides DDE_INPUT_LOCATION as its output, but the app should map that to its own namespace (DOCUMENTAI_INPUT_LOCATION).