Reading
Notes and thoughts from books, papers, and articles I've been reading.
What Does Your Eval Actually Measure?
State of RL for Reasoning LLMs
FlashAttention-4 from Scratch to Near-SOTA in 60 Diagrams (CUDA/PTX)
Extending Raschka's GPT-2: an MoE trained from scratch on an RTX 3090
ML Job Interviews: The Ultimate Guide
Reinforcement Learning from Verifiable Rewards
DeepSeek-V4: Annotated Paper Walkthrough
How to Enter Side Doors
The Big Thing in 2030? World Models
What Happens When You Run a CUDA Kernel
Deep Dive into Yann LeCun's JEPA
Dreaming Is the New Thinking
The Harness Playbook
Simulating the Visual World with Artificial Intelligence: A Roadmap
Mini-LSM: Learn Database Internals
Frontier Model Training Methodologies
How to Optimize a CUDA Matmul Kernel for cuBLAS-Like Performance: A Worklog
How I Use LLMs as a Staff Engineer in 2026
Learnings from MLSys '26
RL Systems Mind the Gap: Matching Trainer and Generator Throughput
Speeding Up GPU Kernels by 38% with a Multi-Agent System
RL-for-LLMs Wiki
Training a Rust 1.5B Coder LM with Reinforcement Learning (GRPO)
RL Post-Training on Macs
Post-Training 101
How to Scale RL
A Systems Engineering Approach to LLM Agents
From Masks to Worlds: A Hitchhiker's Guide to World Models
Keep the Tokens Flowing: Lessons from 16 Open-Source RL Libraries
Kimi Likes Causal Decision Theory More After RL in Twin Prisoner's Dilemmas
World Models
How to Eval AI Agents — The 2026 Guide
How to Ask for Help from People Who Don't Know You
An Advanced System Architecture Breakdown of OpenAI's Jalapeno Inference Accelerator
Writing Speed-of-Light Flash Attention for 5090 in CUDA C++
Intro to GPUs For the Researcher
Modern GPU Programming for MLSys
No readings match that search or filter.