Everything From This Publication To Help You Level Up In AI

Your complete guide to all the resources in this publication.

Nov 04, 2025

👋 Hello there!

I’m Dr. Ashish Bamania, a London-based software engineer and emergency physician.

I’ve previously been a CTO of a health tech startup, building tools that helped people access better primary healthcare.

Alongside this, I’ve been a top Technology, Data Science & AI writer on Medium, with my content read over 1.5 million times.

I’ve also written multiple technology-focused books that make complex technical concepts accessible through visual explanations and concise, no-fluff content.

I am currently an editor of two fast-growing Substack newsletters, “Into AI” and “Into Quantum,” that make developments in artificial intelligence and quantum computing accessible to a broad audience.

That’s a brief introduction to what I do. What follows is all my work, compiled in this publication over the past several years.

I hope you find it helpful! 😊

LLMs: Architecture & Fundamentals

You Don’t Need Normalization In Transformers Anymore: https://intoai.pub/dynamic-tanh
Revisiting The Basics: Rotary Position Embeddings (RoPE): https://intoai.pub/revisiting-the-basics-rotary-position
Deep-Diving & Decoding The Secrets That Make DeepSeek So Good: https://intoai.pub/multi-head-latent-attention-is-the
LLMs: Common terms explained, simply: https://intoai.pub/llms-common-terms-explained-simply
Meta’s Large Concept Models (LCMs) Are Here To Challenge And Redefine LLMs: https://intoai.pub/metas-large-concept-models-lcms-are
Memory Layers Are Supercharging LLMs Like Never Before: https://intoai.pub/memory-layers-are-supercharging-llms
Superfast Matrix-Multiplication-Free LLMs Are Finally Here: https://intoai.pub/superfast-matrix-multiplication-free
Here Is Google DeepMind’s New Research To Build Massive LLMs With A Mixture Of Million Experts: https://intoai.pub/here-is-google-deepminds-new-research
Chain-of-Draft (CoD) Is The New King Of Prompting Techniques: https://intoai.pub/chain-of-draft-cod-is-the-new-king
An LLM With A Visual Sketchpad Can Now Smash Its Competitors Without One (Even GPT-4o): https://intoai.pub/an-llm-with-a-visual-sketchpad-can
‘SpreadsheetLLM’ Finally Lets LLMs Master Spreadsheets Better Than Ever: https://intoai.pub/spreadsheetllm-finally-lets-llms
A Simple Principle From Noise-Cancelling Headphones Supercharges Transformers Like Never Before: https://intoai.pub/a-simple-principle-from-noise-cancelling
Top 4 Decoding Strategies In LLMs Explained Simply: https://intoai.pub/decoding-strategies-in-llms
The Hidden Optimization Every LLM Uses: https://intoai.pub/llm-optimizations

Reasoning & Advanced LLM Capabilities

Tiny Recursive Model (TRM): A Deep Dive: https://intoai.pub/tiny-recursive-model
Hierarchical Reasoning Model: A Deep Dive: https://intoai.pub/hierarchical-reasoning-model
R-Zero: A Method For Training Reasoning LLMs With Zero Data Is Here: https://intoai.pub/r-zero
DeepSeek-R1 & OpenAI’s o1 Aren’t Still As Intelligent As Portrayed.: https://intoai.pub/deepseek-r1-and-openais-o1-arent
DeepSeek-R1 Beats OpenAI’s o1, Revealing All Its Training Secrets Out In The Open: https://intoai.pub/deepseek-r1-beats-openais-o1-revealing
Apple’s New Research Shows That LLM Reasoning Is Completely Broken: https://intoai.pub/apples-new-research-shows-that-llm
LLMs Fail Embarrassingly At Newer Ways Of Testing Understanding: https://intoai.pub/llms-cannot-truly-understand
LLMs Don’t Really Understand Math.: https://intoai.pub/llms-dont-really-understand-math
You Don’t Need ‘Thinking’ In LLMs To Reason Better: https://intoai.pub/you-dont-need-thinking-in-llms
The Open Source “Agentic Reasoning” Beats Google Gemini Deep Research: https://intoai.pub/the-open-source-agentic-reasoning

Reinforcement Learning & Training Techniques

Reinforcement Learning On Pre-Training Data Improves LLMs Like Never Before: https://intoai.pub/reinforcement-learning-on-pre-training
Proximal SFT: SFT Supercharged By RL Is Here: https://intoai.pub/proximal-sft
LLMs Can Now Be Pre-Trained Using Pure Reinforcement Learning: https://intoai.pub/reinforcement-pretraining
LLMs Can Now Self-Evolve At Test Time Using Reinforcement Learning: https://intoai.pub/test-time-reinforcement-learning
A Detailed Guide To Reinforcement Learning From Human Feedback (RLHF) From Scratch: https://intoai.pub/a-detailed-guide-to-reinforcement
A Detailed History of Optimizers (And How The New ‘Adam-mini’ Optimizer Works): https://intoai.pub/the-new-adam-mini-optimizer-is-here
You Don’t Need Backpropagation To Train Neural Networks Anymore: https://intoai.pub/you-dont-need-backpropagation-to

RAG, Embeddings & Information Retrieval

Learn Hypothetical Document Embeddings (HyDE) By Coding It From Scratch: https://intoai.pub/hyde
Cache-Augmented Generation (CAG) Is Here To Replace RAG: https://intoai.pub/cache-augmented-generation-cag-is
Build A Vector Database From Scratch To Understand RAG In Depth: https://intoai.pub/build-a-vector-database-from-scratch
Building Your First Agentic RAG From Scratch: https://intoai.pub/building-your-first-agentic-rag-from
Your Vector Databases Aren’t Safe Anymore: https://intoai.pub/your-vector-databases-arent-safe
A Human Brain-Inspired RAG Approach Has Reached the New State-of-the-Art: https://intoai.pub/a-human-brain-inspired-rag-approach
‘MedGraphRAG’ Is A Complete Game Changer For AI In Medicine: https://intoai.pub/medgraphrag-is-a-complete-game-changer
Google’s New Algorithms Just Made Searching Vector Databases Faster Than Ever: https://intoai.pub/googles-new-algorithms-just-made

AI Agents & Multi-Agent Systems

Claude Skills: Everything That You Need To Get Started: https://intoai.pub/claude-skills
Model Context Protocol Isn’t Confusing If You Understand It This Way: https://intoai.pub/model-context-protocol
Building A LinkedIn Content Creation Agency Using Google’s Agent Development Kit: https://intoai.pub/building-a-linkedin-content-creation
Building Your First AI Agent (That Will Actually Improve You As An AI Engineer): https://intoai.pub/building-your-first-ai-agent
A Squad Of Open-Source LLMs Can Now Beat OpenAI’s Closed-Source GPT-4o: https://intoai.pub/a-squad-of-open-source-llms-can-now

Medical & Healthcare AI

Delphi-2M Predicts Which Disease Will Strike You In the Next 20 Years: https://intoai.pub/delphi-2m
Microsoft Reveals Its Path To Medical Superintelligence: https://intoai.pub/microsoft-path-to-medical-superintelligence
HuatuoGPT-o1: A Medical Reasoning LLM From China With OpenAI o1 Like Capabilities Is Here: https://intoai.pub/huatuogpt-o1-a-medical-reasoning
Is OpenAI’s o1 the AI Doctor We’ve Always Been Waiting For? (Surprisingly, Yes!): https://intoai.pub/is-openais-o1-the-ai-doctor-weve
‘ReXplain’ Is Transforming Radiology With AI Like Never Before: https://intoai.pub/rexplain-is-transforming-radiology
AI Can Now Look At Your Heart & Predict How Long You’ll Live.: https://intoai.pub/ai-can-now-look-at-your-heart-and
AI Is Now Treating Cancer (And This Begins An Exciting Era For Modern Medicine): https://intoai.pub/ai-is-now-treating-cancer-and-this
AI Is Discovering New Antibiotics – A Task Previously Deemed Extremely Challenging For Humans.: https://intoai.pub/ai-is-discovering-new-antibiotics
Your First Hands-On Lesson On Using AlphaFold: https://intoai.pub/your-first-hands-on-lesson-on-using
The Algorithm That (Literally) Helps Doctors Save Thousands Of Patients Everyday: https://intoai.pub/the-algorithm-that-literally-helps

Computer Vision & Multimodal AI

Vision Transformers: Straight From The Original Research Paper: https://intoai.pub/vision-transformers
Top Vision Models Cannot Really See Our World: https://intoai.pub/top-vision-models-cannot-see-our-world
‘Transfusion’ Is Supercharging Training Multi-Modal LLMs Like Never Before: https://intoai.pub/transfusion-is-supercharging-training
Google’s New AI Can Hallucinate An Entire Video Game In Real Time Without A Game Engine: https://intoai.pub/googles-new-ai-can-hallucinate-an
‘Skeleton Recall Loss’ Is The New Breakthrough In Segmentation: https://intoai.pub/skeleton-recall-loss-is-the-new-breakthrough
Stable Signature: Meta’s Solution To End Misinformation Spread With AI-Generated Images: https://intoai.pub/stable-signature-metas-solution-to

Neural Network Architectures & Innovations

Kolmogorov-Arnold Networks (KANs) Might Change AI As We Know It, Forever: https://intoai.pub/kolmogorov-arnold-networks-kans-might
Kolmogorov-Arnold Networks (KANs) Are Being Used To Boost Graph Deep Learning Like Never Before: https://intoai.pub/kolmogorov-arnold-networks-kans-are
XNets Are Here To Beat MLPs & KANs (& How Can You Train One From Scratch): https://intoai.pub/what-are-complexnets-and-how-to-train
Fourier Analysis Networks (FANs) Are Here To Break Barriers In AI: https://intoai.pub/fourier-analysis-networks-fans-are
RNNs Are Coming Back To Take Over Transformers (Yes, For Real!): https://intoai.pub/rnns-are-coming-back-to-take-over
‘FANformer’ Is The New Game-Changing Architecture For LLMs: https://intoai.pub/fanformer-is-the-new-game-changing
Deep Dive Into The Mind Of Autoencoders & Visualize Their Internals: https://intoai.pub/deep-dive-into-the-mind-of-autoencoders

LLM Fine-Tuning & Optimization

A Practical Guide To Fast Fine-Tuning Your LLMs With Unsloth: https://intoai.pub/a-practical-guide-to-fast-fine-tuning
Learn To Train A Reasoning Model From Scratch Using GRPO: https://intoai.pub/train-a-reasoning-model-using-grpo
We Have Finally Found A Solution To An Extremely Energy Efficient AI: https://intoai.pub/we-have-finally-found-a-solution

LLM Security & Safety

‘MathPrompt’ Embarassingly Jailbreaks All LLMs Available On The Market Today: https://intoai.pub/mathprompt-embarassingly-jailbreaks

LLMs Pass On Generational Trauma Just Like Humans Do: https://intoai.pub/llms-pass-on-generational-trauma
The World’s First AI Regulation Act Is Finally Here: https://intoai.pub/the-worlds-first-ai-regulation-act

Self-Improving & Evolutionary AI

Darwin-Gödel Machine: The First Self-Improving AI System Is Here: https://intoai.pub/darwin-godel-machine-the-first-self

AI Engineering & Best Practices

10 Must-Know Concepts For AI Engineers To Build Better Systems: https://intoai.pub/10-must-know-concepts-for-ai-engineers
Tech Shouldn’t Be This Hard To Learn: https://intoai.pub/tech-shouldnt-be-this-hard-to-learn
What Is Class Imbalance In Machine Learning & How To Fix It: https://intoai.pub/what-is-class-imbalance-in-machine
5 Extremely Useful Plots For Data Scientists That You Never Knew Existed: https://intoai.pub/5-extremely-useful-plots-for-data

Federated Learning

A Hands-On Guide To Federated Machine Learning With ‘Flower’: https://intoai.pub/a-hands-on-guide-to-federated-machine

Data Structures, Algorithms & Systems Design

5 Unpopular Tree Data Structures To Level Up As A Software Developer: https://intoai.pub/5-unpopular-tree-data-structures
5 Not-So-Popular Data Structures From Redis (That Will Make You A Better Engineer): https://intoai.pub/5-not-so-popular-data-structures
5 Data Structures That You Probably Are Unfamiliar With (But Are Extremely Useful!): https://intoai.pub/5-data-structures-that-you-probably
6 Mind-Blowing Uses Of Bitwise Operations That You Probably Never Knew Existed: https://intoai.pub/6-mind-blowing-uses-of-bitwise-operations
How Twitter Generates Scalable Unique IDs For Its Tweets: https://intoai.pub/how-twitter-generates-scalable-unique

Python Programming

This Is How You Make Your Python Functions 5000% Faster With Rust (Yes, You Read It Right!): https://intoai.pub/this-is-how-you-make-your-python
Sorry Python Programmers But You Are Missing Out On These 5 Astonishing Rust Features!: https://intoai.pub/sorry-python-programmers-but-you
5 Matplotlib Methods That You Never Knew Existed: https://intoai.pub/5-matplotlib-methods-that-you-never
Hacking Python Internals: Learn To Add A New Keyword To The Python Language: https://intoai.pub/hacking-python-internals-learn-to
Understanding GIL & How It Hinders Scaling Python Applications: https://intoai.pub/understanding-gil-and-how-it-hinders
0.1 + 0.2 is not 0.3 in Python. Here Is Why!: https://intoai.pub/01-02-is-not-03-in-python-here-is

Rust Programming

Modern Ransomware Is Being Written In Rust. (Bad News - It Is Extremely Resilient.): https://intoai.pub/modern-ransomware-is-being-written
Rust Is Getting Extremely Popular. Here Are Some Important Projects Using It. (Even The Windows Kernel!): https://intoai.pub/rust-is-getting-extremely-popular
Google Grants $1 Million To Move From C++ To Rust. Is C++ Slowly Dying?: https://intoai.pub/google-grants-1-million-to-move-from
Bend - A Rust-Backed Language That Supercharges Parallel Computing Like Never Before: https://intoai.pub/bend-a-rust-backed-language-that

JavaScript Programming

JavaScript’s Event Loop & Callback Queue (A Graphical Guide): https://intoai.pub/javascripts-event-loop-and-callback

Systems & Databases

Let’s Peek Inside The Database That You Use Everyday: https://intoai.pub/lets-peek-inside-the-database-that
Row vs. Column-Oriented Formats – This Is What You Really Need To Know Before Making A Choice.: https://intoai.pub/row-vs-column-oriented-formats-this
Ditch JSON! Here Are 5 (Better) Data Serialization Formats To Use In Your Next Project: https://intoai.pub/ditch-json-here-are-5-better-data
A Deep Dive Into How Docker Really Works: https://intoai.pub/a-deep-dive-into-how-docker-really
6 Scheduling Algorithms That Developers Must Know About: https://intoai.pub/6-scheduling-algorithms-that-developers
Multi-Threading Is Not Always The Right Answer. Here’s Why!: https://intoai.pub/maybe-multi-threading-is-not-always

Cryptography & Security

Following Apple and Signal, Zoom Now Moves To Post-Quantum Cryptography. Is Classical Cryptography Dying?: https://intoai.pub/following-apple-and-signal-zoom-now
Here Is How Modern Encryption Works (Visually Explained!): https://intoai.pub/here-is-how-modern-encryption-works
A Deep Dive into SHA-256 By Learning To Hash by Hand: https://intoai.pub/a-deep-dive-into-sha-256-by-learning
How Google Authenticator Works Under The Hood: https://intoai.pub/how-google-authenticator-works-under
Implementing A Pseudorandom Number Generator In Python From Scratch: https://intoai.pub/implementing-a-pseudorandom-number
7 Ways To Store Passwords In Your Database: Level Noob to Level Boss: https://intoai.pub/7-ways-to-store-passwords-in-your
How Apple Pay Works Under The Hood: https://intoai.pub/how-apple-pay-works-under-the-hood
How To Generate A Bitcoin Wallet Address From Scratch: https://intoai.pub/how-to-generate-a-bitcoin-wallet
TailsOS: An Operating System I Would Use If I Wanted To Disappear From The World: https://intoai.pub/tailsos-an-operating-system-i-would

Developer Tools & Git

5 Git Concepts To Move Beyond The Beginner Developer: https://intoai.pub/5-git-concepts-to-move-beyond-the
5 Bash Commands That Sometimes Even Hackers Confuse With (But You Don’t Have To!): https://intoai.pub/5-bash-commands-that-sometimes-even

Interview Prep & Coding Challenges

3 LeetCode One-Liners That Look Easy (But Can Make One P*ss Their Pants In A Tech Interview): https://intoai.pub/3-leetcode-one-liners-that-look-easy
Google Rejected Max Howell (Creator of Homebrew) For Getting This Interview Question Wrong. Can You Solve It?: https://intoai.pub/google-rejected-max-howell-creator
ML Interview Essentials: Building Self-Attention From Scratch: https://intoai.pub/self-attention

Web & UX Design

Use These 5 Psychological Hacks To Drastically Increase Clicks On Your Webpage! (Apply Wisely): https://intoai.pub/use-these-5-psychological-hacks-to

AI Applications & Real-World Use Cases

AI Is Breaking Past Human Discoveries. This Is How It Just Made Sorting Algorithms Faster Than Ever!: https://intoai.pub/ai-is-breaking-past-human-discoveries
AI Just Made Data Compression Algorithms Multiple Times Better Than Ever: https://intoai.pub/ai-just-made-data-compression-algorithms
AI Just Made YouTube Video Compression More Efficient Than Ever. Here Is How!: https://intoai.pub/ai-just-made-youtube-video-compression
Hold Tight, Nuclear Fusion Is Coming Sooner Than You Think (Thanks To Artificial Intelligence!): https://intoai.pub/hold-tight-nuclear-fusion-is-coming

Philosophy & Game Theory

A Computer Science Proof Of Why Nice Guys Do Not Finish Last (But First).: https://intoai.pub/a-computer-science-proof-of-why-nice

Books & Resources

“LLMs In 100 Images” Is Now Out!: https://intoai.pub/llms-in-100-images-is-here
🏆 100 images >> 1000 research papers: https://intoai.pub/100-images-1000-research-papers
My Latest Book “Computer Science In 100 Images” Is Out!: https://intoai.pub/my-latest-book-computer-science-in
The ‘Tech Visualised’ Series Is Out & Growing!: https://intoai.pub/the-tech-visualised-series-is-out
‘AI In 100 Images’: I’m Making My New Book Available For Free: https://intoai.pub/ai-in-100-images-im-making-my-new
Grab A 30% Discount On My Book “Systems Design In 100 Images”: https://intoai.pub/grab-a-30-discount-on-my-book-systems
I’m Writing a Book on Reinforcement Learning: https://intoai.pub/im-writing-a-book-on-reinforcement

📩 Subscribe to ‘Into AI’ — a newsletter that helps you become a better engineer, by exploring the latest and best in Artificial Intelligence from the ground up.

Check out my books on Gumroad and connect with me on LinkedIn to stay in touch.

Into AI