How to Deploy Your First Serverless Function on AWS Lambda

Quick answer

• An AWS Lambda function is code plus a trigger; the simplest trigger to start with is a Function URL.
• Write a handler that takes an event and returns a response, upload it, and test with a URL.
• Success is an HTTP 200 response and a matching entry in CloudWatch Logs.

Step 1 — Write the function

Create a file lambda_function.py:

import json

def lambda_handler(event, context):
    name = event.get("queryStringParameters", {}).get("name", "world")
    return {
        "statusCode": 200,
        "headers": {"Content-Type": "application/json"},
        "body": json.dumps({"message": f"Hello, {name}!"})
    }

The handler receives an event (the request data) and returns a response object.

Step 2 — Create the function in the console

In the AWS Lambda console, choose Create function → Author from scratch, name it hello-function, and choose the Python runtime. Paste the code above into the editor and click Deploy.

How to verify it worked: the editor saves without errors and shows “Changes deployed.”

Step 3 — Add a Function URL

In the function’s Configuration → Function URL, enable it with Auth type: NONE for a public test endpoint, then copy the generated URL.

Step 4 — Invoke the function

curl "https://YOUR-ID.lambda-url.REGION.on.aws/?name=Ada"

How to verify it worked: the response is {"message": "Hello, Ada!"} with an HTTP 200.

Step 5 — Read the logs

Add a print statement, redeploy, and invoke again. In the Lambda console’s Monitor → View logs in CloudWatch, you’ll see the invocation with your print output. How to verify it worked: each invocation produces a log stream entry, proving observability is working.

Step 6 — Add a secret properly

Don’t paste an API key into the code. Add it in Configuration → Environment variables as API_KEY, then read it in the handler with os.environ["API_KEY"].

Where this bites vibecoders

The first Lambda an AI assistant generates often works — then it’s left with a public Function URL and no auth, or a secret hardcoded in the source. Treat “it returned 200” as step one, not the end: lock down the endpoint and move secrets into environment variables before calling anything done.

Where AI coding assistants get this wrong

• Hardcoding secrets in the handler instead of using environment variables.
• Leaving a public Function URL with Auth type: NONE on anything non-trivial.
• Ignoring timeout and memory, so long work gets killed with a generic error.
• Writing a handler that can’t parse its own event shape and fails on the first real request.

Checklist

• Keep the handler small and single-purpose.
• Test the real event shape, not a made-up one.
• Store secrets in environment variables, never in code.
• Set timeout and memory to match the workload.
• Confirm logs are flowing before you rely on the function.

FAQ

What is a Lambda handler?

A handler is the entry-point function the Lambda runtime calls, receiving an event and a context and returning a response. Its name is configured in the function’s runtime settings (here, lambda_function.lambda_handler).

How much does Lambda cost?

Lambda bills by invocations and compute time (GB-seconds), with a generous free tier. A rarely-called function costs essentially nothing; a high-traffic one can add up. Set up billing alerts so costs stay visible.

When should I not use Lambda?

For long-running processes (a function has a maximum timeout), latency-critical endpoints sensitive to cold starts, or steady always-on workloads where a server is cheaper. See What Is Serverless Computing?.

Related topics

• How to Deploy Your First App on AWS for Free
• What Is Serverless Computing?
• How to Manage Secrets and Environment Variables Properly
• What Is FinOps?

Sources

• AWS Lambda documentation
• AWS Lambda Function URLs