
Weekend Creative Agent Challenge: CodeMentorAI
CodeMentorAI is an autonomous AI coding interview coach built with AWS. It uses Amazon Bedrock and Nova Lite to generate fresh daily coding challenges. EventBridge, Lambda, S3, and SES automate generation, storage, and email delivery. Built to turn interview preparation from a habit into an automated routine
«“Don’t wait for motivation. Automate consistency.”»
Preparing for coding interviews is not only about knowing Data Structures and Algorithms.
The bigger challenge is staying consistent.
As a student balancing academics, projects, and interview preparation, I often found myself postponing practice. I had to remember to open a platform, search for a problem, choose a difficulty, and finally start practicing.
That made me think:
«What if my interview preparation could start before I even opened the application?»
That question became the idea behind CodeMentorAI.
🚀 The Idea
CodeMentorAI is an always-on AI coding interview coach that autonomously creates a fresh coding interview challenge every day using Generative AI.
Each challenge contains:
- A Data Structures & Algorithms interview question
- Difficulty level
- A helpful hint
- An interview preparation tip
The content is generated using Amazon Bedrock and Amazon Nova Lite.
But the most important part isn't simply the AI generation.
The user doesn't have to initiate it.
There is no Generate button.
There is no manual Lambda execution.
There is no need to open an application first.
The system works automatically in the background according to a configured schedule.
When I check my inbox, the day's coding challenge is already waiting.
That is the problem CodeMentorAI is designed to solve:
«Turning interview preparation from something I have to remember into something that happens automatically.»
🧠 Why an Autonomous Approach?
A traditional AI application usually depends on a user initiating every interaction.
I wanted to approach the problem differently.
Instead of building another application that waits for the user, I wanted to build a system that could prepare something for the user automatically.
This made CodeMentorAI an opportunity to explore:
- Event-driven architecture
- Serverless computing
- Generative AI
- Automated content generation
- Cloud storage
- Automated communication
- IAM and service-to-service permissions
The AI model is only one part of the system.
The surrounding cloud infrastructure is what makes the workflow autonomous.
🏗️ How I Built CodeMentorAI
Amazon EventBridge Scheduler
The workflow starts with Amazon EventBridge Scheduler.
It automatically triggers the workflow according to the configured schedule, removing the need for manual execution.
This scheduled trigger is the key component that gives CodeMentorAI its always-on behavior.
AWS Lambda
I built the orchestration layer using Python on AWS Lambda.
Lambda coordinates the entire process.
It receives the scheduled invocation, prepares the structured prompt, invokes Amazon Bedrock, processes the generated response, stores the challenge, and handles the email delivery workflow.
Amazon Bedrock + Amazon Nova Lite
I integrated Amazon Bedrock to provide the Generative AI capability.
The project uses Amazon Nova Lite to generate the daily coding interview challenge.
I designed a structured prompt so that the model produces concise and useful interview preparation content, including the question, difficulty, hint, and preparation tip.
Amazon S3
After generation, Lambda stores the challenge in Amazon S3.
Each challenge is saved as a timestamped JSON object.
This provides persistent storage and creates a history of the challenges generated by the system.
Amazon SES
The generated challenge is then delivered using Amazon SES.
I formatted the content as an HTML email so the user receives the coding challenge directly in the inbox without needing to open an application.
AWS IAM
AWS IAM manages the permissions required between the services.
The Lambda execution role is configured with the permissions needed to interact with services such as Amazon Bedrock, Amazon S3, and Amazon SES.
🛠️ Challenges I Faced
Building the project also meant dealing with real cloud infrastructure problems.
IAM Permission Issue
One of the first issues I encountered occurred when Lambda attempted to upload the generated challenge to S3.
The Lambda execution role did not initially have the required:
"s3:PutObject"
permission.
After reviewing the Lambda execution logs, I identified the missing permission, updated the IAM policy, and tested the workflow again.
Once the permission was correctly configured, Lambda successfully stored the generated challenge in S3.
Amazon SES Sandbox
I also encountered an issue with email delivery because the AWS account was operating in the SES Sandbox environment.
Email sending was restricted because the required identities had not yet been verified.
I configured the necessary verified identities and tested the workflow again.
After verification, the system successfully delivered the generated HTML email.
These challenges taught me an important lesson:
«An AI model alone does not make an application autonomous.»
Scheduling, permissions, storage, delivery, logging, and service integration all have to work together.
🤖 Demonstrating the Autonomous Agent
The autonomous behavior of CodeMentorAI can be demonstrated through the AWS services involved in the workflow.
The evidence includes:
- EventBridge Scheduler showing the configured schedule
- Lambda logs showing scheduled executions
- Amazon Bedrock generating the challenge
- Amazon S3 containing newly generated timestamped JSON objects
- Amazon SES delivering the generated challenge through email
Together, these demonstrate that the system is not dependent on a user manually clicking a button.
The creative output is generated, stored, and delivered automatically.
That is what makes CodeMentorAI different from a traditional AI application.
📚 What I Learned
Building CodeMentorAI gave me practical experience with:
- AWS Lambda
- Amazon EventBridge Scheduler
- Amazon Bedrock
- Amazon Nova Lite
- Amazon S3
- Amazon SES
- AWS IAM
- Python
- Serverless architecture
- Event-driven architecture
- Generative AI integration
- Cloud permissions and troubleshooting
More importantly, the project changed how I think about AI applications.
Previously, I mostly viewed an AI application as:
Input → AI → Output
This project showed me that an autonomous AI system can involve much more:
Event → AI Generation → Storage → Delivery → User
The AI model is only one component.
The real value comes from combining the model with cloud infrastructure that allows it to operate automatically.
🎯 Why This Project Matters
CodeMentorAI started with a very simple problem:
«How can I make coding interview preparation more consistent without depending entirely on motivation?»
Instead of building another tool that waits for the user to return, I wanted to build something that works while the user is away.
Every scheduled execution creates a new challenge.
The challenge is stored in the cloud.
The result is delivered directly to the user.
And the entire process happens without requiring a button click.
💡 The Biggest Lesson
The biggest lesson I took from this project is:
«“The best automation isn't the one that saves seconds — it's the one that removes the need to remember.”»
For interview preparation, that means I don't have to remember to search for a new problem every day.
The system creates the opportunity automatically.
☁️ AWS Services Used
This project brought together several AWS services:
Amazon EventBridge Scheduler — automated scheduling and triggering
AWS Lambda — serverless Python orchestration
Amazon Bedrock — Generative AI integration
Amazon Nova Lite — AI foundation model for challenge generation
Amazon S3 — persistent storage for generated challenges
Amazon SES — automated HTML email delivery
AWS IAM — secure service-to-service permissions
Together, these services form the infrastructure behind CodeMentorAI.
🙌 Final Thoughts
CodeMentorAI began with a simple idea:
What if technology could help me stay consistent before motivation disappears?
That idea turned into a serverless, event-driven Generative AI workflow built with AWS.
Through this project, I explored how Amazon Bedrock, Amazon Nova Lite, AWS Lambda, EventBridge Scheduler, Amazon S3, Amazon SES, and IAM can work together to create an always-on AI system.
I didn't want to build technology that simply waits for the user to return.
I wanted to build technology that works while the user is away.
«CodeMentorAI
Don't wait for motivation. Automate consistency.»
🔗 Project & Connect
#AWS #AWSBuilderCenter #AWSCloud #AmazonBedrock #AmazonNova #GenerativeAI #Serverless #AWSLambda #AmazonS3 #AmazonSES #EventBridge #CloudComputing #AI #Python #CodingInterviews #Developer #StudentDeveloper #agents
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article