AWS Builder Center
AWS builder portal

AWS builder portal

AWS builder portal on Nimbus Hackathon

AWS Student Builder Groups Club Member Portal — AI-Powered Knowledge & Community Assistant

Introduction

Student communities often share important information through multiple documents, chat groups, announcements, and messages. Finding the right information at the right time can become difficult, especially during workshops, hackathons, and other events.
To address this challenge, we built the AWS Student Builder Groups Club Member Portal, an AI-powered web application that provides authenticated members with a centralized platform to access club information and interact with a document-grounded chatbot.
The portal allows members to sign up, log in, reset their password, explore club documents, and ask questions through an AI chatbot. Instead of generating answers from general knowledge, the chatbot retrieves information from the club's approved documents and provides the relevant source along with the answer.
For the second phase of the challenge, we extended the original portal with an Admin Document Management and Knowledge Synchronization system. Club administrators can now publish or update documents, trigger re-indexing, and make newly published information available to the chatbot and member dashboard within approximately 60 seconds.
The project was designed to be fully demonstrable locally while also providing a clear path for deployment on AWS.



What We Built

The Club Member Portal combines:
  • Secure member authentication
  • Password reset functionality
  • Member dashboard
  • AI-powered document-based chatbot
  • Retrieval-Augmented Generation (RAG)
  • Source citations
  • Admin document management
  • Document publishing and updating
  • Automatic re-indexing
  • Knowledge synchronization
  • Real-time/simple dashboard updates
  • Smoke testing
  • API access through  POST /ask 
The application is designed around one important principle:
If the information is not available in the club documents, the chatbot should not guess.
Instead, it provides a helpful fallback directing the member to the campus AWS Student Builder contact.
This follows the project's requirement to answer only from club documents and show sources on every reply.



Who This Helps

The portal is primarily designed for AWS Student Builder Group members and club administrators.

For Members

A new or existing member can:
  • Create an account
  • Log in securely
  • Reset their password
  • Access the member dashboard
  • View available club documents
  • Ask questions through the chatbot
  • Receive answers based on club documentation
  • See the source used for each answer
  • Access newly published information without manually refreshing the application
For example, instead of searching through multiple documents for information about a workshop or AWS Builder Center, a member can simply ask the chatbot a question.



For Club Administrators

Administrators can:
  • Access a protected admin dashboard
  • Add new documents
  • Upload or paste document content
  • Update existing documents
  • Publish documents
  • Trigger re-indexing
  • Monitor indexing status
  • Make new information available to the chatbot
  • Keep the member document list updated
This makes the portal useful not only as a chatbot but also as a continuously updated knowledge platform.



Core Features

1. Member Registration and Login

Members can create an account using their email and password.
After authentication, members can access the protected dashboard and chatbot.
The portal does not require social login.
The authentication flow includes:
  • Sign up
  • Login
  • Logout
  • Forgot password
  • Password reset
The baseline specification requires email/password authentication and a password-reset flow.



2. Member Dashboard

The dashboard provides members with a central place to access the club's knowledge base.
It displays:
  • Welcome information
  • Available documents
  • Recently published documents
  • Document status
  • Last updated information
  • Knowledge-base updates
When an administrator publishes a new document, the dashboard can update the document list without requiring a complete browser reload.



3. AI-Powered Chatbot

The chatbot is the main interaction point for members.
Members can ask natural-language questions about:
  • AWS Student Builder Groups
  • Workshops
  • Builder Center
  • AWS services covered by the documents
  • Hackathon rules
  • Community information
  • Other information contained in the club knowledge base
The chatbot uses the indexed documents as its knowledge source.
Instead of relying purely on general AI knowledge, the system follows a retrieval-based approach:
Question → Search documents → Retrieve relevant content → Generate grounded answer → Display sources



4. Retrieval-Augmented Generation (RAG)

The project uses a document-grounded RAG approach.
The knowledge base contains the approved club documents.
The initial starter pack contains:
  •  01-onboarding-faq.md 
  •  02-aws-account-setup.md 
  •  03-builder-center-publish.md 
  •  04-bedrock-starter.md 
  •  05-hackathon-rules.md 
  •  06-workshop-index.md 
  •  07-lambda-patterns.md 
  •  08-sbg-community.md 
These documents form the initial knowledge base for the chatbot.
When a member asks a question, the system searches the knowledge base for relevant content and uses the retrieved information to generate the answer.



5. Source Citations

One of the most important features of the portal is source transparency.
Every chatbot response should show the document used to generate the answer.
For example:
Answer: The answer retrieved from the relevant club documentation.
Source:  06-workshop-index.md 
Depending on the retrieval implementation, the system can also expose:
  • Document name
  • Chunk ID
  • Rank
  • Relevance score
This makes the chatbot easier to trust and verify.
The project requirements explicitly require sources to be shown on every answer.



6. Safe AI Fallback

The chatbot is designed not to hallucinate information.
If a question cannot be answered from the available documents, the system responds with a fallback such as:
"I could not find that in the club documents. Please contact the campus AWS Student Builder contact."
This is important because the application is intended to provide reliable information from an organization's approved knowledge base rather than unrestricted AI-generated answers.
The baseline also explicitly instructs the system not to invent AWS pricing, limits, or policies.



7. Admin Document Management

The second phase extends the original portal with an administrator workflow.
Admins can:
  1. Open the Admin Dashboard
  1. Add a document
  1. Upload or paste content
  1. Publish the document
  1. Trigger indexing
  1. Monitor indexing status
  1. Make the document searchable
This means administrators can update the knowledge base without rebuilding the entire application.



8. Automatic Re-indexing

The most important addition in the 30% phase is knowledge synchronization.
When an administrator publishes a document:
Publish → Process → Chunk → Index → Update Knowledge Base → Chatbot Uses New Content
The target synchronization time is approximately 60 seconds.
The requirement specifically states that newly published content should become searchable within about a minute.



9. Chat Synchronization

The chatbot does not require the member to refresh their browser after an administrator publishes a document.
For example:
An administrator publishes an updated event document containing a new room number.
Within the synchronization window, a member can ask:
"Where is the event happening?"
The chatbot retrieves the updated information and cites the newly published document.
This demonstrates that the chatbot is connected to the latest knowledge index rather than relying on a static dataset loaded only when the application starts.



10. Event-Day Briefing

The application supports the event-day document:
 event-day-briefing.md 
This document contains information such as:
  • Today's schedule
  • Rooms
  • Judging times
  • Deadlines
The chatbot must be able to retrieve information from this document and cite it as the source.
This feature is particularly useful during a live event because schedules and locations can change quickly.



11. Smoke Testing

We also added a smoke-test workflow for validating the knowledge system.
The application can execute the three required questions from:
 09-smoke-test-questions.md 
For each question, the system can show:
  • Question
  • Expected source
  • Retrieved answer
  • Retrieved source
  • PASS / FAIL
This gives the team a simple way to demonstrate that the RAG system is retrieving the correct information.
The hackathon requirements specifically include demonstrating the three smoke-test questions.



12. API Endpoint

The application exposes a required API endpoint:
POST /ask

The server runs on:
0.0.0.0:8080

The request format is:
{
"question": "..."
}

The response contains the answer and sources at the top level:
{
"answer": "The grounded answer...",
"sources": [
{
"document": "event-day-briefing.md",
"chunk_id": "optional-team-chunk-id",
"rank": 1,
"score": 0.91
}
]
}

The evaluator reads  answer  and  sources  directly from the response, so the response is not wrapped inside another object such as  data  or  result .



13. Why This Approach?

Traditional club portals usually require users to manually search through documents.
A general-purpose AI chatbot can answer questions quickly, but it may provide information that is outside the organization's approved documents.
Our approach combines both ideas:
Club Portal + Document Search + RAG + Source Citations
This provides:
  • Easier information discovery
  • Faster answers
  • Source transparency
  • Controlled knowledge
  • Easier document updates
  • Better support for changing event information
The administrator can update the knowledge base without rebuilding the entire application.



Technology and Architecture

The application is designed around a web frontend, backend API, document storage, search/indexing layer, and AI/RAG layer.
A conceptual architecture is:
┌──────────────────────┐
│ Member │
│ Web Application │
└──────────┬───────────┘
│
▼
┌──────────────────────┐
│ Chat Interface │
└──────────┬───────────┘
│
▼
┌──────────────────────┐
│ /ask API │
└──────────┬───────────┘
│
▼
┌──────────────────────┐
│ Retrieval / RAG │
└──────────┬───────────┘
│
┌─────────────┴─────────────┐
▼ ▼
┌─────────────────┐ ┌─────────────────┐
│ Document Index │ │ AI Model │
│ / Vector Search │ │ Amazon Bedrock │
└────────┬────────┘ └────────┬────────┘
│ │
└────────────┬─────────────┘
▼
Grounded Answer
+
Sources




AWS Deployment Architecture

Although the application can be demonstrated locally, the project has a clear AWS deployment path.
The baseline specification suggests an AWS architecture using services such as Amazon Cognito, Amazon SES, Amazon S3, the chat API, and AI services.
A possible cloud architecture is:

Authentication

Amazon Cognito
Used for:
  • User registration
  • Login
  • Authentication
  • Password management

Email

Amazon SES
Used for:
  • Password reset emails
  • Authentication-related email communication

Document Storage

Amazon S3
Used to store:
  • Club documents
  • Newly uploaded documents
  • Updated versions

Processing

AWS Lambda
Used to:
  • Process uploaded documents
  • Split documents into chunks
  • Generate embeddings
  • Trigger indexing

Search

Amazon OpenSearch
Used as the searchable/vector knowledge layer.

AI

Amazon Bedrock
Used for generating answers based on retrieved document content.

API

Amazon API Gateway + AWS Lambda
Can expose the chatbot API in a production AWS architecture.



Document Publishing Flow on AWS

The cloud version could work as follows:
Admin
↓
Upload / Publish Document
↓
Amazon S3
↓
AWS Lambda
↓
Document Processing
↓
Chunking + Embeddings
↓
Amazon OpenSearch
↓
Updated Knowledge Index
↓
Amazon Bedrock
↓
Grounded Chatbot Answer
↓
Source Citation

This architecture would allow the knowledge base to scale while keeping the chatbot grounded in approved documents.
Live AWS deployment is optional for the challenge; the official requirements state that a local demo is sufficient, while live AWS deployment can earn bonus marks.



Security and Responsible AI

We designed the system with several safety principles:

Document-grounded answers

The chatbot should answer only from available club documents.

No guessing

If relevant information cannot be retrieved, the system uses the fallback instead of inventing an answer.

Source transparency

Every answer includes its source.

Authentication

Members must authenticate before accessing the chat.

Role-based access

Administrative functions are restricted to admins.

Sensitive data protection

The system should not log passwords or private student information.



Demo Flow

Our recommended demonstration flow is:

1. Member Registration

Show the signup page and create a member account.

2. Login

Log into the member portal.

3. Ask a Question

Ask a question related to the starter documents.

4. Show Source

Demonstrate that the answer contains the source document.

5. Test Unknown Information

Ask something that is not available in the knowledge base.
Show the safe fallback.

6. Admin Login

Open the administrator dashboard.

7. Publish a Document

Add or update  event-day-briefing.md .

8. Re-index

Show:
Publishing → Indexing → Indexed

9. Dashboard Synchronization

Show the newly published document appearing on the member dashboard without a full page reload.

10. Chat Synchronization

Ask a question about the newly published content.
The chatbot should answer using the new document.

11. Source Verification

Show:
Source: event-day-briefing.md

12. Smoke Tests

Run all three smoke tests and show their PASS/FAIL results.

13. API Test

Call:
POST http://<machine-ip>:8080/ask

and show the required JSON response.
This demo flow directly covers the expected judging flow for the 30% extension.



What Makes the Project Different?

The main idea is not simply "an AI chatbot."
It is an AI-powered, document-grounded community portal with live knowledge synchronization.
The important workflow is:
Admin updates knowledge → system re-indexes → members automatically receive updated information → chatbot answers from the latest documents → sources are shown.
This is particularly useful for student communities where information such as schedules, rooms, workshops, deadlines, and announcements can change frequently.



Future Improvements

If the project is extended beyond the hackathon, possible improvements include:
  • Full AWS deployment
  • Amazon Cognito integration
  • Amazon S3 document storage
  • Amazon Bedrock integration
  • Amazon OpenSearch vector search
  • Automated document processing
  • Push notifications
  • Advanced analytics
  • Admin audit logs
  • Document version history
  • Multi-club support
  • Improved semantic search
  • Mobile application
  • Multilingual support
These are future possibilities and are not required for the current local hackathon implementation.



Conclusion

The AWS Student Builder Groups Club Member Portal brings authentication, community knowledge, document search, AI assistance, source citations, and live knowledge updates into one platform.
The 70% baseline provides the foundation with member authentication and document-grounded chat. The 30% extension adds the critical ability for administrators to publish new information and synchronize the chatbot and dashboard with the latest knowledge.
Our goal is to make information access within student communities faster, safer, more transparent, and easier to maintain.
The project demonstrates how Retrieval-Augmented Generation can be combined with a practical community portal to create an AI assistant that is not only conversational, but also grounded in an organization's trusted knowledge base.



Project Highlights

Project: AWS Student Builder Groups Club Member Portal
Core concept: Document-grounded AI Community Assistant
Key technologies/concepts:
  • Web Application
  • Authentication
  • RAG
  • Document Search
  • AI Chatbot
  • Source Citations
  • Knowledge Synchronization
  • Document Indexing
  • REST API
  • Amazon Bedrock
  • AWS Architecture
Required API:
 POST /ask 
Port:
 8080 
Primary focus:
Trusted answers from trusted documents, with continuously updated knowledge.



Recommended Screenshots for the Builder Center Article

Add screenshots in this order:
  1. Landing / Home Page
  1. Sign Up Page
  1. Login Page
  1. Forgot Password Page
  1. Member Dashboard
  1. Chatbot with Answer + Source
  1. Admin Dashboard
  1. Add/Upload Document Page
  1. Indexing Status
  1. Updated Member Document List
  1. Chat Answer Using Newly Published Document
  1. Smoke Test PASS/FAIL Screen
  1. POST  /ask  API Response
  1. AWS Architecture Diagram
The baseline specifically recommends screenshots showing signup, login, forgot password, and chat with sources in the Builder Center article.

Tags

 #aws-student-builders-groups   #buildonaws   #amazon-bedrock   #rag 
Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article