Textractify: AI-Powered Receipt Processing in the Cloud - Part 2

#aws #sideprojects #ai #cloudproject

🛠️ Step-by-Step Implementation: Automating Receipt Processing Using AWS

This guide walks you through the full setup of a serverless receipt processing pipeline using AWS services.

1️⃣ Create the S3 Bucket (for uploading receipts)

✅ Steps:

Go to the S3 Console → Click Create Bucket
Name your bucket (e.g., storage-receipt-omkarsharma2821)
Choose a region (e.g., ap-south-1)
Click Create bucket
Create Organizational folder inside bucket.
Name it incoming inside this you will upload files.

2️⃣ Create a DynamoDB Table (to store extracted data)

✅ Steps:

Go to the DynamoDB Console → Click Create Table
Table name: Receipts-table
Partition key: receipt_id (String)
Sort-key: date (String)
Click Create

3️⃣ Set Up Amazon SES (to send emails)

✅ Steps:

Go to Amazon SES Console
Verify your sender email under Verified Identities
(Optional) Verify recipient email if your account is in sandbox mode
Note the region (e.g., ap-south-1) – you’ll need it in your Lambda

4️⃣ Create IAM Role for Lambda Execution

✅ Steps:

Go to the IAM Console → Roles → Create Role
Choose Lambda as the use case
Attach the following policies:

AmazonS3ReadOnlyAccess
AmazonTextractFullAccess
AmazonDynamoDBFullAccess
AmazonSESFullAccess
AWSLambdaBasicExecutionRole
1. Name the role: LambdaReceiptProcessingRole

5️⃣ Create Lambda Function (processing engine)

✅ Steps:

Go to AWS Lambda Console → Click Create Function
Name: ProcessReceiptFunction
Runtime: Python 3.9 or Node.js
Choose existing role → Select LambdaReceiptProcessingRole
Go to configuration tab inside environment varibales add this.
Go to the Code tab and add the pyhton code that I provide in python.py file and click Deploy.
Go to configuration tab > General configuration > edit
Increase the timeout from 0.3 sec to 2 min for complex file.

6️⃣ Again go to S3 Bucket.

✅ Steps:

In the Properties Tab
Add the Event Notification
Prefix : incoming/
Object creation : Select All object create events

Wait for 30 sec and also check in spam folder for the mail....if you do not receive mail after 2 min go to the monitor tab in Lambda Function and check the log groups in cloudwatch. you can verify in dynamodb table as well below are the attached proofs