MeldIt

End-to-end automation platform for Youngun's meme-marketing operation — from ad-campaign registration to creator distribution to verification of what creators actually posted, plus payments and analytics. The verification layer used Instagram scraping, video segmentation, and image/sub-section matching to confirm posts at scale.

Status
Shipped
Client
Youngun
Context
Meme-marketing agency operations platform — scaled what used to be Google Forms + manual verification.
Role
Built the first version end to end — image-matching pipeline, Instagram scraping, payment flow, content distribution. Handed off to a frontend lead who turned it into the full app.

Bill of materials

  • Python
  • Image/video matching
  • Instagram scraping
  • Media caching
  • Payment distribution

Youngun runs meme-marketing campaigns at scale — they own a network of meme pages and work with a roster of creators who post sponsored content into the feed. Before MeldIt, the operations layer was almost entirely manual: a creator would post their piece, paste the link into a Google Form, someone on the team would manually open it, eyeball whether the content matched the brief, copy numbers into a spreadsheet for analytics, and queue the payment. Every campaign meant hundreds of these checks.

MeldIt automates the entire loop.

What it does

  • Campaign registration — brief in, media specifications captured, distributed to the right meme pages and creators.
  • Distribution — routes the right brief to the right creators based on the campaign’s needs.
  • Verification at scale — the hard part. Pulls posted content from Instagram, runs video segmentation and image / sub-section matching against the brief’s reference media, and confirms whether what got posted is what was supposed to get posted. Not LLM-based — actual visual matching, because “did this creator post the thing we paid them to post” needs to be unambiguous.
  • Analytics — engagement, reach, performance, all aggregated automatically instead of by hand.
  • Payment distribution — once verification clears, the payout flows.

What I built

I created the first version end to end — the matching pipeline, large-scale Instagram scraping, media caching, the data plumbing between every stage, and the operational scaffolding around it. I then handed it off to the frontend lead who turned it into the full application Youngun runs today.