FintechAI Services

Intelligent Loan Processing Platform

Loan document intake for a US lender handling thousands of documents a month. The contract was extended 7 times.

Key Details

Challenge	A US fintech lender funding SMB loans and equipment purchases needed to take in and check large volumes of loan documents.
Solution	An ASP.NET and React platform with Azure ingestion, Form Recognizer extraction, a validation engine and bureau and banking integrations.
Technologies	ASP.NET Core, React, Azure Functions, Azure Blob, Azure AD B2C, Azure DevOps, Azure Form Recognizer

Technologies used

ASP.NET CoreReactAzure FunctionsAzure BlobAzure AD B2CAzure DevOpsAzure Form Recognizer

Client

The client is a fast-growing US fintech lender working on SMB loans and equipment financing. The platform takes documents in, extracts fields and validates them against connected data sources.

Product screenshots

Light teal document review screen with a scanned bank statement on the left showing highlighted extracted regions, and on the right the extracted fields with confidence values, validation checks that pass or need review, and a document checklist for the loan file. — Loan document intake: an uploaded bank statement is read by OCR, fields are pulled into the application and each one is checked against the rest of the file.

Dark blue architecture diagram of an asynchronous document pipeline: upload sources feed blob storage and a message queue, serverless workers run classification and OCR extraction, a validation engine checks results against bank and credit bureau connections, and results flow to an underwriter review app, with live throughput and queue-depth tiles. — Ingestion pipeline: documents arrive through the portal or email, are queued and processed asynchronously, read by OCR, validated against bureau and bank data, and land in the underwriter's queue.

What we built

ASP.NET and React platform
Asynchronous document ingestion on Azure
OCR extraction with Azure Form Recognizer
Validation engine
Integrations with Plaid, Equifax and Experian

Engagement

Delivery was time and materials with 8 engineers. The initial term was 3 months. It was extended 7 times and ran for 23 months.

Project team: 8 engineers. Duration: 23 months. Model: Time & materials.

Outcome

7Contract extensions

ThousandsDocuments a month

23 moTotal engagement

Services used

AI Services →Workflow AI Agent →

Industries

Fintech →