The Codex is the long-form part of the site: 17 study paths with 476 pages between them. Each path starts with one long page that takes a subject from the basics to what you need in production, with the books, papers and courses worth reading, and goes on to shorter pages that look at one part in detail.
What every program runs on: the processor, the operating system and the network.
16 pages
How a CPU runs your code: pipelines, branch prediction, caches, virtual memory, NUMA, SSDs and GPUs, with figures from current chips.
Architecture internals One page per part of the machine, from the instruction cycle to boot.11 pages
Processes, threads, scheduling, virtual memory, file systems, inter-process communication and system calls.
Operating system internals Processes, memory, the page cache, IPC and syscalls in detail.13 pages
From sockets and IP through TCP, QUIC, TLS, DNS, BGP and NAT, to the problems that show up in production.
The networking stack One page per layer, from byte order to routing between networks.28 pages
Pages organised by the question you have, such as what is using the CPU, which process has a port open, or why the disk is full.
When a server is slow The first commands to run when all you know is that something is slow.Storing data safely, and keeping copies of it in agreement across machines.
15 pages
B-trees, LSM trees, MVCC, isolation levels, write-ahead logs and replication, and how real databases combine them.
Database internals Storage engines, transactions, isolation and the query planner.21 pages
Replication, consensus, clocks and partial failure: the problems every system spread across machines has to deal with.
Distributed systems topics Replication, consensus, time, idempotence, quorums, leases and failure detection.Designing a service, its interfaces, and keeping it fast, observable and secure.
27 pages
Estimating capacity, choosing storage, replication, background work and multi-region setups, with worked designs such as a URL shortener, a news feed and a chat service.
The design playbook Interview designs from the first clarifying questions to failure modes.12 pages
REST, gRPC, GraphQL, Protobuf, WebSockets and webhooks, plus versioning, authentication and idempotency.
Protocols and wire formats One page per protocol or format, with the trade-offs between them.9 pages
Ways to find out why something is slow: the USE and RED methods, latency percentiles, queueing theory, profiling and load testing.
Methods Each method with the tools it uses and the mistakes people make with it.4 pages
Logs, metrics and traces and when to use each, OpenTelemetry, SLOs and error budgets, and eBPF.
Logs, metrics and traces What each signal is good for, and a worked incident that uses all three.5 pages
Threat modelling, secrets management, authentication, the common classes of vulnerability, and the cryptography an engineer needs day to day.
Security pages Threat models, secrets, sign-in, vulnerability classes and cryptography.Where services run, and the newer kind of service built around language models.
15 pages
From your first kubectl commands to running clusters and preparing for the CKA, with the books, courses and labs worth your time.
Kubernetes internals The control plane, the data plane, what happens on kubectl apply, and controllers.45 pages
Compute, storage, networking, identity, managed databases, multi-region and cost, on AWS with the GCP and Azure equivalents.
Cloud topics One page per area, built around patterns rather than certification exams.5 pages
How AI features run in production: tokens and embeddings, transformers, inference and the KV cache, vector search, RAG and agents.
How LLMs work Tokens, embeddings, attention and decoding, one step at a time.Writing code well in a specific language, and solving problems under interview conditions.
96 pages
Go, Rust and JavaScript, each as a long study page plus pages on how its runtime works.
Go internals The scheduler, channels, the garbage collector, escape analysis, cgo and reflection.128 pages
Twenty patterns that cover most interview questions, such as two pointers, sliding windows, BFS, dynamic programming and union-find, plus worked problems for each.
The pattern playbook How to recognise each pattern, a reference solution, and problems from easy to hard.7 pages
The low-level design round: problems such as an LRU cache, a parking lot and a vending machine, with classes, state and edge cases.
The problems Requirements, entities, interfaces, state machines and concurrency for each.