Martin Andrews

@martinandrews-mdda · 1,670 subscribers · 40,557 views · 17 videos · SG · community-listed, not yet claimed · Watch on YouTube

About

Welcome to The Mechanics of Modern AI. This channel is for AI builders, developers, and engineers who need to understand how modern AI systems really work.

Are you an "AI Architect" who's past the introductory tutorials and hitting the limits of treating AI as a black box? You're in the right place. We deconstruct the foundational mechanics behind today's most important models, covering topics like LLM reasoning and prompt engineering from a first-principles, research-driven perspective. My mission is to empower you to stop just using AI and start truly understanding it.

Subscribe to get the under-the-hood insights you need to debug, innovate, and build more robust applications with confidence.

Boards

#101 in Best AI & ML Channel

Recent uploads

Using signals between the tokens = Next-level LLM RL training
Use your old GPU to run a modern LLM!
Become a Model Whisperer : The "On-Policy" Secret to Better LLM results
World Models & Neural Assets: The Mechanics of AI Simulation
How an AI Agent Won Gold at the Physics Olympiad (Paper Explained)
How Mixture of Experts (MoE) Actually Works
Do LLMs Know When They're Wrong?
Hacking an LLM's Personality with Representation Engineering
Latent Space Reasoning : Looking at the research
AlphaEvolve and Darwin Gödel Machines : LLMs for Code Evolution in 2025
AI That Evolves Its Own Prompts
Beyond Optimization: The Power of AI Diversity

Around it on Best AI & ML Channel

#98 SystemDR - Scalable System Design · 2,620 subscribers
#99 Machine learning classroom · 2,310 subscribers
#100 TechOnline with Peace Tutorials · 2,170 subscribers
#102 Padh Le Yaar · 1,560 subscribers
#103 School of Machine Learning · 1,220 subscribers
#104 The JavaScript Way · 1,210 subscribers
#105 The Researchers' Guide · 1,120 subscribers

Positions on paid boards are paid placement. Listing here does not imply the creator's participation or endorsement unless marked ownership verified.

Rank on the paid boards is paid placement. The Earned board cannot be bought.