I'm building Mimestra — it finds viral short-form videos,
breaks down why they went viral, and turns that into a shoot-ready brief. This post is
about one infra decision that saved the project.

The cheap path that bit me

For Instagram data I went the obvious cheap route: a burner IG account + instagrapi,
self-hosted. Great in testing. Then, right as I onboarded first users, the account got
suspended. No data, broken product, worst timing.

If you've scraped IG yourself you know the tax: sessions, 2FA, proxies, rate limits —
you babysit infrastructure instead of building the product. And the ban isn't "if",
it's "when".

Collapsing it into one call

I moved the IG layer to HikerAPI (a managed Instagram REST API —
they run the account pool, proxies and sessions). My whole multi-step "collect" stage
turned into a single request:

python
import requests

r = requests.get(
    "https://api.hikerapi.com/v1/media/by/code",
    params={"code": "SHORTCODE"},
    headers={"x-access-key": "YOUR_KEY"},
)
media = r.json()
# one response → author, metrics, pk, AND a direct video_url
print(media["video_url"], media["like_count"])