Inspiration
One fine day, my girlfriend and me started to discuss about what to cook based on the ingredients, time and utensils available. About 6 hours passed and we ended up ordering outside. Thats when I though, although we have google search, what if we had an application that could develop recipe based on special preferences like healthy version of fast food with low calories.
What it does
It takes centered inputs from users through hosted webpage on streamlit, plugs the inputs into refining the default prompt that provides instructions for formatted output on the backend. Next, it sends an api call to Gemini AI with prompt and displays the response.
How we built it
I went though some tutorial on using api key to get response from LLM. Later watched tutorial on developing and hosting web application using Stremlit. Lastly combined best of both with Gemini AI (as it was free to use) and created the web application
Challenges we ran into
Primary challenge was installing gemini modules on top of all existing modules on my systems took most of the time. Rest all was simple and straight forward
Accomplishments that we're proud of
Proud of building a customized version of LLM for food recipe based on prompt engineering and hosting it publicly on Streamlit
What we learned
I learned how Gemini api can be used as processing engine for generating results faster without worrying about storing information into a database. Hosting web application Stremlit
What's next for Get Food Recipe
Next steps will be fine tuning prompt, creating user profile to add persona for the prompt, saving recipes. Granting ability to produce images along with prompt response. I have received some review about taking image as input and produce result, might implement the flow.


Log in or sign up for Devpost to join the conversation.