FindAThing

Inspiration

Fiddling with filters and engineering your queries is a problem that should be left in the 2010's. With natural language becoming more popular, it's time to do away with old search engine style language, and let our unique language find us the best products.

What it does

Uses natural language, mixed with 'classic' search filters, to find the perfect product for you from Amazon's millions of products.

How I built it

LlamaIndex, Hugging Face and NiceGUI; I tend to stick with Python.

Challenges I ran into

• Finding and engineering a good dataset of Amazon products.
• Building a simple, but good looking UI.
• Prompt engineering.

Accomplishments that I'm proud of

• Getting a RAG product ready

What I learned

• How to use LlamaIndex with a local vector database
• Building a vector database

What's next for FindAThing

• Supporting more sources of product data (e.g., Etsy)
• Supporting more input formats (e.g., image, voice)

Built With

• openai
• python