Astera Support Desk

Astera Support Desk is a customer-service chatbot for ecommerce support teams. It is designed to handle common customer queries such as order tracking, delivery delays, refunds, payment issues, address changes, and account-access problems through a polished web interface and a lightweight NLP-style intent engine.

This project was built to demonstrate practical product thinking, backend API design, frontend execution, and applied NLP fundamentals in a way that feels closer to a real support tool than a classroom prototype.

Live Demo

Open Astera Support Desk

Portfolio Summary

Astera Support Desk is a polished full-stack chatbot project that demonstrates:

• applied NLP fundamentals through intent detection and fuzzy matching
• backend API design with Flask
• frontend product thinking through a branded, responsive support interface
• realistic business use cases in ecommerce customer support

This is the kind of project that is easy to discuss in interviews because it combines technical implementation with clear product value.

Live Repository Preview

Why This Project Stands Out

• Solves a real business problem by reducing repetitive customer-support workload
• Combines backend logic, UI design, and conversational experience in one project
• Uses explainable intent detection instead of a black-box model, making it easy to discuss in interviews
• Presents the chatbot as a believable product with clean branding and realistic support flows
• Shows readiness for future scaling into live APIs, CRM systems, and ML-based classification

Product Overview

Astera Support Desk acts as a first-line digital support associate. A customer can open the chat interface, ask a natural question, and receive immediate guidance for common support cases.

Examples:

• "Where is my order?"
• "My payment failed but money was deducted."
• "Can I change my delivery address?"
• "My coupon is not working."
• "I forgot my password."

The chatbot identifies the most likely user intent, returns a relevant answer, and suggests useful follow-up questions to keep the conversation moving.

Core Features

• Rule-based chatbot with fuzzy intent matching
• Responsive support-desk style frontend
• Flask backend with a dedicated /chat API endpoint
• Intent detection using keyword overlap and string similarity
• Support for realistic ecommerce issues such as refunds, payment failures, delayed delivery, and account help
• Quick action suggestions for smoother user interaction
• Graceful fallback responses for unsupported questions

Supported Customer Intents

• Greeting
• Order tracking
• Shipping information
• Delivery delays
• Refund policy
• Return process
• Order cancellation
• Payment methods
• Payment failures
• Account help
• Address change
• Product availability
• Promo code issues
• Store hours
• Contact support
• Goodbye

Tech Stack

• Python
• Flask
• HTML
• CSS
• JavaScript
• difflib.SequenceMatcher for lightweight fuzzy matching

Architecture Snapshot

System Design

1. Frontend

The frontend provides a modern support-chat experience with:

• branded hero section
• conversational chat layout
• avatar-based message styling
• quick-action suggestion chips
• responsive design for desktop and mobile

2. Backend

The Flask server:

• serves the main UI
• accepts chat messages through a JSON API
• routes user messages to the chatbot engine
• returns structured responses with reply text, detected intent, confidence score, and follow-up suggestions

3. Chatbot Engine

The chatbot engine:

• normalizes user input
• compares it against predefined intent patterns
• scores matches using keyword overlap and fuzzy similarity
• selects the best intent above a confidence threshold
• falls back gracefully when no strong match is found

Project Structure

Astera Support Desk/
|-- app.py
|-- chatbot_engine.py
|-- requirements.txt
|-- README.md
|-- static/
|   |-- style.css
|   `-- script.js
|-- templates/
    `-- index.html

What Recruiters Can Look At

Product Thinking

• The chatbot is framed as a usable support experience, not just a technical demo
• The interface and conversation design were shaped around realistic customer needs

Engineering Skills

• Clean separation between UI, API, and chatbot logic
• Structured JSON responses between frontend and backend
• Intent matching logic that is simple, explainable, and extensible

UX and Frontend Execution

• Visually polished interface
• Smooth chat interaction with suggestions and typing feedback
• Mobile-friendly responsive layout

Extensibility

• Easy to connect to order databases or courier APIs
• Easy to replace the rule-based engine with an ML or LLM-based classifier later

GitHub About Section

If you want to use a short description in the GitHub repository "About" field, use:

Polished Flask-based customer support chatbot with rule-based NLP, fuzzy intent matching, and a responsive ecommerce help desk UI.

Local Setup

1. Create a virtual environment

python -m venv .venv

2. Activate the environment

.venv\Scripts\activate

3. Install dependencies

pip install -r requirements.txt

4. Run the application

python app.py

5. Open in browser

http://127.0.0.1:5000

Free Deployment Option

As of April 17, 2026, the easiest free option for this Flask project is PythonAnywhere.

Why this is the best fit:

• PythonAnywhere still offers a free Beginner account
• it supports Flask directly
• it is simpler for this project than converting the app to serverless hosting
• it does not require the gunicorn startup flow used by platforms like Render

Important free-plan limits:

• 1 web app
• 1 web worker
• 512 MiB disk space
• 2 consoles
• 1 month expiry on the free web app

Official sources:

• PythonAnywhere free account features: https://help.pythonanywhere.com/pages/FreeAccountsFeatures
• PythonAnywhere pricing: https://www.pythonanywhere.com/pricing/
• Flask setup guide: https://help.pythonanywhere.com/pages/Flask

Deploy To PythonAnywhere

1. Create a free PythonAnywhere account.
2. Open the Web tab.
3. Click Add a new web app.
4. Choose Manual configuration and select Python 3.
5. Upload or clone this GitHub repo into your PythonAnywhere home folder.
6. Open the WSGI configuration file from the Web tab.
7. Replace its contents with the code from pythonanywhere_wsgi.py, but change:

project_home = Path("/home/yourusername/Chatbot")

to your real PythonAnywhere username and folder path.

8. Reload the web app from the Web tab.

Notes

• On PythonAnywhere, your Flask app is loaded through the WSGI file, not by running python app.py.
• The if __name__ == "__main__": block in app.py is fine and will not run during PythonAnywhere import.
• If you want, I can next simplify the repo for PythonAnywhere specifically and remove the Render-only files.

Example Questions To Test

• Where is my order?
• My order is delayed.
• I want a refund.
• My payment failed but money was deducted.
• Can I change my delivery address?
• My coupon code is not working.
• I forgot my password.
• How can I contact support?

Future Improvements

• Add real order-tracking integration
• Add customer authentication and profile history
• Store chat logs or support tickets in a database
• Add analytics for common support issues
• Upgrade intent detection with machine learning or LLM-based routing
• Deploy the app to Render, Railway, or another cloud platform

Interview Talking Points

• Why a rule-based NLP system can be a strong first version for support automation
• How fuzzy matching improves intent detection over exact keyword matching
• Why support-chat UX matters as much as backend accuracy
• How this architecture can evolve into a production-ready support assistant

License

This project is open for educational and portfolio use.