---
title: "Reconstructed Slides: From MCP to Scale: Pipelines That Build Themselves — Rafael Levi, Bright Data"
category: "slides"
video_id: "zTZ0qunQXnM"
sourceLabels: ["Cropped public YouTube video frames", "Local OpenCV slide-region detection", "Local RapidOCR"]
---

# Reconstructed Slides: From MCP to Scale: Pipelines That Build Themselves — Rafael Levi, Bright Data

## Source Video
[From MCP to Scale: Pipelines That Build Themselves — Rafael Levi, Bright Data](https://www.youtube.com/watch?v=zTZ0qunQXnM)

## Method
This deck is reconstructed from the existing video frame captures by detecting likely slide regions with OpenCV, cropping/upscaling those regions, deduplicating similar crops, and OCRing the cropped slide images locally. It is a cleaner companion to the full-stage frame deck.

## Reconstructed Slides
![[assets/reconstructed-slides/zTZ0qunQXnM/slide-007.jpg]]

- Recreated text/layout view: [open HTML recreation](/assets/slide-recreations/reconstructed/zTZ0qunQXnM/slide-007.html)
- AI slide classifier: `content_slide` confidence `0.94`
- Text source: advanced OCR `rapidocr-live/left-72/contrast`.
- OCR decision: ready — Dense code editor screenshot projected on screen; OCR is better than manual transcription.

Slide text:

> 14 11 Leportos LinportSsom
> Laperttim
> 88 11 12 29 laport Froebilepvtotntifulsoo lapertloering rejests
> ...Configiratioe
> 24
> AIE 24 otrT_o·.Nh1mm（o.t._1_))
> 27
> ★ ★ 24 sog·logting.getlegter ring.nicContig（lewel.oeglag.two,fornst.（octiee）sXeoisago）
> 31 Jnt (otopogo(wlt str, mtries) Lnt. 3) "fetopsge Uroopteipt wrawet mlecter (orsttaegtnrege（retries)) Fetcner ery? （
> "attpa://ops.trigitdata.con/regont
> 月 Jenf°roe';zont,'oel*!sr]. lratien fer
> 41 reipose.roiu_fee_sutn)
> Ntwre.resoonse.text
> 6.8 coiselonioetrrer(rieledtetet(s(or)ofter(retries)Mteets 1attept（rotries·11 lme.alo()*.atteng)
> AI Enginee
> EUROPE

![[assets/reconstructed-slides/zTZ0qunQXnM/slide-008.jpg]]

- Recreated text/layout view: [open HTML recreation](/assets/slide-recreations/reconstructed/zTZ0qunQXnM/slide-008.html)
- AI slide classifier: `content_slide` confidence `0.93`
- Text source: advanced OCR `rapidocr-live/border-trim/contrast`.
- OCR decision: ready — Dense product/UI screenshot with tables and small text; OCR is better than manual transcription.

Slide text:

> Sarch
> ? chedied
> Tideor
> Corteat
> 500
> RnngGoacraen g
> AIE
> n
> leLeM
> DN 4922V L4PV
> ibi AI Engineer
> EUROPE


### Hidden Non-Slide Evidence
- [`slide-001.jpg`](/assets/reconstructed-slides/zTZ0qunQXnM/slide-001.jpg) — `sponsor_logo` confidence `0.99`; Sponsor wall, not a presentation slide.
- [`slide-002.jpg`](/assets/reconstructed-slides/zTZ0qunQXnM/slide-002.jpg) — `speaker_stage` confidence `0.97`; Presenter shot with projected demo UI and stage branding, not a standalone slide.
- [`slide-003.jpg`](/assets/reconstructed-slides/zTZ0qunQXnM/slide-003.jpg) — `speaker_stage` confidence `0.98`; Presenter shot with app UI on the screen, not a standalone slide.
- [`slide-004.jpg`](/assets/reconstructed-slides/zTZ0qunQXnM/slide-004.jpg) — `speaker_stage` confidence `0.98`; Presenter shot with app UI on the screen, not a standalone slide.
- [`slide-005.jpg`](/assets/reconstructed-slides/zTZ0qunQXnM/slide-005.jpg) — `demo_video` confidence `0.95`; Live browser demo frame, not a presentation slide.
- [`slide-006.jpg`](/assets/reconstructed-slides/zTZ0qunQXnM/slide-006.jpg) — `speaker_stage` confidence `0.98`; Presenter shot with browser demo on stage, not a standalone slide.
- [`slide-009.jpg`](/assets/reconstructed-slides/zTZ0qunQXnM/slide-009.jpg) — `speaker_stage` confidence `0.98`; Stage photo with speaker and audience; projected slide is not the primary frame.
- [`slide-010.jpg`](/assets/reconstructed-slides/zTZ0qunQXnM/slide-010.jpg) — `title_card` confidence `0.99`; Branding/title card with no substantive presentation content.

Classification audit: `raw/sources/slide-ai-classification/reconstructed/zTZ0qunQXnM/audit.json`