Skip to content

fix: reorder case studies by impact + rewrite scraping challenges - #3

Merged
rafaelortizdotdev merged 4 commits into
mainfrom
feat/chartmetric-case-studies
Aug 27, 2026
Merged

rafaelortizdotdev merged 4 commits into
mainfrom
feat/chartmetric-case-studies

Conversation

@rafaelortizdotdev

Copy link
Copy Markdown
Owner

Summary

  • Reordered all case studies by impact — Chartmetric (1-3), US Retail (4), HONOR (5), Tag Insight (6)
  • Rewrote scraping case study with real challenge descriptions per platform:
    • Ticketmaster: considered impossible, made it work
    • Discogs: 7 months unsolved by another engineer (coordination/math problem)
    • TikTok: total outage recovery, $1,500→$50/mo
    • Tunefind: API+HTML alignment at scale
    • Shazam: millions of requests, cost discipline
    • Genius: precise throttle balancing
  • Redesigned ScrapingReach component — challenge-type cards with results instead of fake difficulty bars
  • Zero implementation details revealed

Test plan

  • Case studies render in correct order on /case-studies and /en/case-studies
  • ScrapingReach component animates correctly
  • Existing case studies still render fine at new order positions

🤖 Generated with Claude Code

…enges

- Chartmetric case studies now order 1-3 (highest impact), existing ones 4-6
- Replaced fake difficulty percentages with real challenge descriptions:
  Ticketmaster (impossible), Discogs (7mo unsolved), TikTok (outage),
  Tunefind (API+HTML), Shazam (scale), Genius (throttle precision)
- Visual component redesigned: challenge-type cards instead of bars
- No implementation details revealed — results only

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
@vercel

vercel Bot commented Aug 27, 2026 •

Copy link
Copy Markdown

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated (UTC)
portfolio-backend Error Error Aug 27, 2026 2:08pm

- Watchdog: removed self-inflicted 71-alert story, reframed as replacing
  months of manual firefighting with automated monitoring (before=chaos, after=stability)
- CH Cloud contract: changed to ongoing negotiation (not completed)
- Added ~$12K/mo target Snowflake savings from Analytics_Daily migration

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Watchdog section: 80% DAG failure rate → <20% across 5 Airflow services
- 60% reduction in automated alerts (data_alert + SF warehouse rearrange)
- Updated results section with both metrics (EN + ES)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Stats bar now shows: 288 migrated, ~$12K target savings, 60% fewer alerts,
<20% DAG failures (was: 288 migrated, $4K/mo, 38 sources)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
@rafaelortizdotdev
rafaelortizdotdev merged commit a77085e into main Aug 27, 2026
1 of 2 checks passed

This branch had an error being deployed

1 failed deployment
Preview — 70820e81 Deployed Aug 27, 2026 by vercel[bot]
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant