HPNVIDIADeveloperChallenge

Inspiration

Shopping for appliances online for purchasing can be overwhelming due to multiple listings and vast amount of choices. We wanted to simplify this by enabling the search process that actually understands what users are looking for — and QuickCart was born.

What it does

QuickCart allows users to search for appliances based on their preferences for purchasing. It returns relevant results by understanding the meaning of queries using a transformer model. It also removes duplicate listings, converts prices to USD, and displays product images, details, and links.

How we built it

• HP AI Studio
• Hugging Face Transformers for embeddings
• MLflow for model deployment
• App is deployed to Swagger through AIS
• Pandas for data processing
• scikit-learn for similarity calculations
• Data source for appliances: https://www.kaggle.com/datasets/lokeshparab/amazon-products-dataset

Challenges we ran into

We had challenges to learn how to set up and use HP AI Studio for the project. We had to read multiple examples and documentations to learn about the steps and concepts.

Accomplishments that we're proud of

We’re proud of creating a functional app that is easy to use. The integration of model deployment, data processing, and a clean user interface was a significant achievement.

What we learned

We learnt to use HP AI Studio and NVIDIA to work with members and collaborate to build the project. We learnt to use models and workspace that the studio offers to scale the project.

What's next for QuickCart

We plan to expand QuickCart beyond appliances to include other retail categories. Future work includes refining the search results using user feedback, adding voice-based queries, and integrating with real-time e-commerce APIs for dynamic pricing and availability.

Built With

• css3
• flask
• hp-ai-studio
• hpnvidiadeveloperchallenge
• html5
• mlflow
• nvidia
• pandas
• python
• scikit-learn