Lytport || (my most complex project so far)

Social media advisor using Machine Learning and stats for data-driven, human-friendly suggestions.

56shots_so.png

link to demo


What it does:

I highly recommend you watch the demo video

Takes all the insight from user's IG pages

adds it to our db

using the already scraped 7500 pieces of content on IG in the Education niche, it gives you the perfect combination of insight, creativity and a comparative analysis

use the chat component to make sense of all the data and get specific questions answers

generate a report based on what you want and get graphs dynamically generated based on the content of the report and your page insight

Why did we create it:

we wanted to see how we could use Bert, Skitlearn to make sense of some data, then we thought what if we scraped IG and then do something with the information. but useful, so we convinced the office of global education to let us use their page and solve the most pressing problem they have.

get insight and understand it, know what other pages are doing and innovate, and then generate report to inform team or higher ups

and that's what we did.

check out the slide deck presentation

Main Techstack:

Nextjs;

Tailwind;

Google SQL Query, Instagram Graph API;

Built a Custom Scraper ( 7.5K content scraped ) ;

Random Forest Regressor;

Google Text to Speech ( for accessibility ) ;

Deployed on GCP

Moudjahid