10 parts
Equation Cheatsheets
Every key equation, annotated. Each term links to a post explaining the why — the intuition, the derivation, and an interactive visual. A reference layer for the blog.
cheatsheetequationsreferencerltransformersmath
Series
Multi-part series that build a topic up from the ground.
Cheatsheets
Every key equation annotated term-by-term — each one links to the post explaining the why, with interactive visuals.
Interactive
Research blogs you can operate — drag the dials and watch the architecture recompute live.
2 parts
The Architecture Playground
Interactive research blogs where you can turn the dials yourself — drag the config of a transformer and watch the weight matrices, KV cache, and expert routing recompute in real time.
architecturetransformersinteractivekv-cachemoe
Defining Frontier
What separates frontier models in each domain — LLMs, VLMs, agents, and beyond.
16 parts
Mechanistic Interpretability
How do transformers actually work inside? From representation geometry and attention circuits to sparse autoencoders, steering vectors, and attribution methods — a rigorous series on understanding what neural networks compute and why.
interpretabilitymechanistic-interpretabilitytransformerscircuitsfeatures
8 parts
Responsible AI and Evaluation
How to design safety benchmarks, detect harmful outputs, build runtime guardrails, and measure what actually matters in production AI systems.
responsible-aisafetyevaluationbenchmarking
5 parts
LLM Evaluation & Benchmarks
A frontier-level survey of how language models are measured — instruction-following, chat quality, creative writing, formatting, and reward models. Per benchmark: leaderboard data, saturation analysis, and what the numbers actually mean.
llm-evaluationbenchmarksleaderboardssaturation
Fundamentals
Ground-up series on the mechanics behind modern AI systems.
12 parts
Reinforcement Learning for LLMs
From policy gradients to RLHF, GRPO, and agentic systems — a ground-up series on how RL shapes modern language models.
rlllm-trainingalignment
6 parts
The LLM Training Lifecycle
From raw text to instruction-following assistant — a stage-by-stage breakdown of how large language models are pre-trained, mid-trained, fine-tuned, and aligned, and what changes at each stage.
llm-trainingpretrainingsftrlhfdpoalignment
14 parts
Transformer Training Internals
The engineering details that determine whether your model converges, generalizes, and runs efficiently — from normalization choices to optimizer internals.
transformersllm-trainingarchitecture
14 parts
Self-Supervised Learning and World Models
From contrastive learning to JEPA and beyond — how machines learn without labels and why it points toward world models.
sslself-supervisedworld-modelsvision
14 parts
Diffusion Models: From Noise to Image
Score matching, DDPM, latent diffusion, DiT, and flow matching — the full arc of how diffusion models became the dominant generative architecture.
diffusiongenerative-modelsimage-generation
22 parts
Mathematics for LLM Training
Linear algebra, probability, optimization, and information theory — the mathematical foundations that every transformer paper builds on.
mathlinear-algebraprobabilityllm-training
6 parts
Ranking and Rank
Two kinds of rank shape modern LLMs: the rank that orders models by preference — Bradley–Terry, ELO, and reward models — and the rank of a matrix — SVD and low-rank adaptation. This series develops both from first principles.
preference-learningsvdlow-rankreward-modelslinear-algebra
16 parts
Inference Systems for LLMs
Prefill vs decode, continuous batching, memory management, and serving at scale — the engineering behind fast and efficient LLM inference.
inferencellmsystemsperformance
4 parts
GPU and Kernel Optimization
From CUDA thread blocks and shared memory to custom Triton kernels, fused operators, and profiling with nsys — the implementation layer behind fast LLM inference.
gpucudatritonkernelsinferenceoptimization
9 parts
Linear Algebra for ML
From vectors and dot products through eigendecomposition and SVD — the mathematical foundations that underpin modern ML systems, built up from geometry with interactive visualizations at every step.
mathlinear-algebrageometryml-foundations