---
title: "Dense Slides: The 2025 AI Engineering Report — Barr Yaron, Amplify"
category: "slides"
video_id: "mQ7_Zje7WKE"
sourceLabels: ["Captured video frames", "Local OpenCV slide-region detection"]
---

# Dense Slides: The 2025 AI Engineering Report — Barr Yaron, Amplify

## Source Video
[The 2025 AI Engineering Report — Barr Yaron, Amplify](https://www.youtube.com/watch?v=mQ7_Zje7WKE)

## Method
This deck is slide-only. The existing captured video frame set supplies candidate frames, then local OpenCV rejects sponsor/title/speaker-only frames, crops visible slide surfaces, deduplicates, and saves the cropped slide images.

## Cropped Visible Slides
![[assets/dense-slides/mQ7_Zje7WKE/slide-002.jpg]]

- Recreated text/layout view: [open HTML recreation](/assets/slide-recreations/dense/mQ7_Zje7WKE/slide-002.html)
- AI slide classifier: `content_slide` confidence `0.99`
- Text source: advanced OCR `rapidocr-live/bright-screen/contrast`.
- OCR decision: ready — dense chart labels and small text

Slide text:

> best describes your work title? Survey demographics: which option respondents identified as engineers-either software or Al. The largest group of
> Woi Software Engincer AI Englneer
> Foundor
> Management
> VP/CEO/CTO
> MLor MLOps Engincor
> Researcher
> Product Manager
> Other
> Data Sciontist
> 6% 10% 15% 20%
> Aroin 02025
> World's Fair AlEngineer Engineering the future of Al

![[assets/dense-slides/mQ7_Zje7WKE/slide-003.jpg]]

- Recreated text/layout view: [open HTML recreation](/assets/slide-recreations/dense/mQ7_Zje7WKE/slide-003.html)
- AI slide classifier: `content_slide` confidence `0.99`
- Text source: advanced OCR `rapidocr-live/bright-screen/opencv-adaptive`.
- OCR decision: ready — dense chart labels and callout text

Slide text:

> What techniques are you using to customize your Ai systems? 70%+ of respondents are using RAG.
> Few-shot learnlng
> RAG
> Flne-tunlng
> Zero-Shot Loarning indicating widespread preference for parameter-efficient methods. 40% mentioned LoRA/QLoRA,
> PretraIning Most popular core training approach is supervised fine-tuning.
> oht
> 10% 20% 301, 601% 70.
> Argty A tagranrg suriy! rrot +

![[assets/dense-slides/mQ7_Zje7WKE/slide-004.jpg]]

- Recreated text/layout view: [open HTML recreation](/assets/slide-recreations/dense/mQ7_Zje7WKE/slide-004.html)
- AI slide classifier: `content_slide` confidence `0.98`
- Text source: advanced OCR `rapidocr-live/bright-screen/contrast`.
- OCR decision: ready — two dense text columns

Slide text:

> Top newsletters from respondents Top podcasts from respondents
> Latent Space (swyx) Latent Space (swyx & Alessio)
> The Batch (Andrew Ng) 62 Machine Learning Street Talk (Tim Scarfe)
> 3 Interconnects (Nathan Lambert) The Cognitive Revolution (Nathan Labenz)
> Ahead of Al (Sebastian Raschka) No Priors (Sarah Guo and Elad Gil)
> AlphaSignal (Lior Alexander) 5 This Week in Machine Learning/Al (Sam C.)
> SemiAnalysis (Dylan Patel) Gradient Dissent (Lukas Biewald)
> Eugene Yan's Newsletter Practical Al (Dan Whitenack)
> Ben's Bites (Ben Tossell) The Gradient (Daniel Bashir)
> Import Al (Jack Clark) Weaviate Podcast (Connor Shorten)
> The Neuron (Noah Edelman) *Robot Brains (Pieter Abbeel)
> AmgtityN taginarg Sunn


### Hidden Non-Slide Evidence
- [`slide-001.jpg`](/assets/dense-slides/mQ7_Zje7WKE/slide-001.jpg) — `speaker_stage` confidence `0.98`; speaker on stage with a partial slide visible, not a standalone presentation slide

Classification audit: `raw/sources/slide-ai-classification/dense/mQ7_Zje7WKE/audit.json`