Serverless CV Processing Pipeline | AWS

A serverless, event-driven pipeline built on AWS that automatically screens candidate CVs and notifies recruiters by email. Built as a hands-on project to apply core AWS services.

---

Architecture

---

How It Works

1. A candidate uploads their CV (file) to an S3 bucket
2. The upload triggers an S3 event notification that invokes a Lambda function
3. Lambda runs keyword matching logic against the CV content
4. Based on the result, Lambda calls SES (Simple Email Service) to send a notification to the recruiter inbox
5. CloudWatch captures all Lambda logs for monitoring and debugging
6. IAM roles grant Lambda the minimum required permissions to access S3 and SES

---

AWS Services Used

Service	Role in this project
S3	Stores uploaded CVs and triggers Lambda on new uploads
Lambda	Runs the keyword matching logic (Python)
SES	Sends the screening result email to the recruiter
CloudWatch	Logs Lambda executions for monitoring and debugging
IAM	Controls permissions between services

---

Lambda Function

The core logic runs in Python. It extracts text from the uploaded CV, checks for a predefined list of keywords, and routes the result to SES.

import boto3
import json
import urllib.parse

s3_client = boto3.client('s3')
ses_client = boto3.client('ses', region_name='us-east-1')

SENDER_EMAIL = 'georgesatef750@gmail.com'
RECIPIENT_EMAIL = 'georgesatef750@gmail.com'

KEYWORDS = ['aws', 'networking', 'python', 'linux', 'cloud', 'cisco', 'ccna']

def lambda_handler(event, context):
    
    bucket_name = event['Records'][0]['s3']['bucket']['name']
    object_key = urllib.parse.unquote_plus(event['Records'][0]['s3']['object']['key'])
    
    print(f"New file uploaded: {object_key} in bucket: {bucket_name}")
    
    response = s3_client.get_object(Bucket=bucket_name, Key=object_key)
    file_content = response['Body'].read().decode('utf-8', errors='ignore').lower()
    
    found_keywords = [kw for kw in KEYWORDS if kw in file_content]
    
    if found_keywords:
        subject = f"Strong candidate: {object_key}"
        body = (
            f"Resume received: {object_key}\n\n"
            f"Keywords detected: {', '.join(found_keywords)}\n\n"
            f"Recommendation: Review this candidate."
        )
    else:
        subject = f"Resume received: {object_key}"
        body = (
            f"Resume received: {object_key}\n\n"
            f"No matching keywords found.\n\n"
            f"Recommendation: Standard review queue."
        )
    
    ses_client.send_email(
        Source=SENDER_EMAIL,
        Destination={'ToAddresses': [RECIPIENT_EMAIL]},
        Message={
            'Subject': {'Data': subject},
            'Body': {'Text': {'Data': body}}
        }
    )
    
    print(f"Email sent. Keywords found: {found_keywords}")
    
    return {
        'statusCode': 200,
        'body': json.dumps('Pipeline executed successfully')
    }

---

Demo

For a full walkthrough with explanation, see the LinkedIn post:

11-minute explained video

---

Screenshots

Step-by-step screenshots are available in the /screenshots folder, covering:

• S3 bucket setup and uploaded CVs
• Lambda function configuration and trigger
• CloudWatch logs after processing strong and weak candidate CVs
• Lambda test results

---

What I Learned

This project pushed me to understand how AWS services communicate with each other, not just what each service does in isolation. Configuring IAM permissions correctly took more iteration than expected and gave me a real appreciation for least-privilege access. Seeing CloudWatch logs reflect exactly what my Lambda code was doing made the event-driven model click in a way that reading about it did not.

---

LinkedIn | GitHub