Here you'll find tutorials on technologies I recently found fascinating, or simply wanted to brush up on.
Tutorials
Sort by
Showing all 34 tutorials, newest first
Python supply chain attacks: how a poisoned release reaches you
Tags:Python
Wheels, startup hooks, how pip and uv resolve versions, lock files and cooldowns, PyPI, CI secrets and publish tokens, release pipelines, signatures and provenance, and what a payload takes, one link at a time, through the two 2026 LiteLLM compromises.
Card payments, part 2: how Visa and Mastercard carry a payment
Tags:Other
Issuer and acquirer processors, access points, the switch and its data centres, risk scores and stand-in, clearing, settlement and disputes, cross-border payments, Visa, Mastercard, Amex, Discover and domestic networks, and where payment data lives, one part at a time, on the same coffee as part 1. Part 2 of 2.
Card payments, part 1: how Apple Pay pays for a coffee
Tags:Other
Card numbers and issuers, acquirers, card networks, authorization, clearing and settlement, interchange, tokens, adding a card to Wallet, the Secure Element, NFC, offline taps and transit gates, and in-app payments, one part at a time, from one tap at a café counter. Part 1 of 2.
Reading htop: how a Linux server is put together
Tags:Other
The /proc files, processes and threads, CPUs and their time modes, system calls and interrupts, the scheduler and load average, virtual memory, the page cache, disks and I/O wait, swap and the OOM killer, cgroups and steal time: every htop number traced to the part of the machine that makes it.
AI agent tracing: how a Phoenix setup is put together
Tags:AI
Spans, OpenInference attributes, trace context, OTLP, Phoenix’s database, evals, prompt versions, experiments, redaction and GDPR erasure, one part at a time, on a shop’s returns agent.
Terraform: how code, state and AWS fit together
Tags:AWSOther
Terraform with the AWS provider, part by part: configuration, providers, plan and apply, the state file, what happens when code, state and AWS disagree, import, moved and removed blocks, lifecycle rules, remote state with locking, and modules, with every plan line from a real run.
OAuth and JWT: how one sign-in becomes an API call
Tags:Other
The authorization endpoint and consent, the token endpoint and PKCE, JWTs and how an API checks them, OpenID Connect, refresh tokens and revocation, the backend-for-frontend pattern and DPoP, one part at a time, from one sign-in to an API call.
Async Python: how one thread checks 100 links
Tags:Python
Sockets and the selector, coroutines, the event loop and futures, tasks and TaskGroup, cancellation and timeouts, semaphores and connection pools, the default thread pool, the GIL, interpreter pools, free-threaded 3.15 and debug mode, built up one part at a time around one link checker.
AWS Bedrock and SageMaker: how an AI support chat is put together
Tags:AWSAI
Bedrock’s Converse API, inference profiles, the accounts Claude runs in, invocation logs, GPU instances, SageMaker AI endpoints, autoscaling, batch, PrivateLink, SCPs and guardrails, added one at a time until a support chat’s whole AI stack fits on one map.
Prometheus: how a monitoring stack is put together
Tags:Other
Client libraries, exporters, scraping, the Pushgateway, the TSDB, PromQL, dashboards, rules and Alertmanager, one part at a time, on a two-copy checkout service, with PCA exam cues.
AWS Route 53: how a company’s DNS is put together
Tags:AWS
Hosted zones, records, routing policies, health checks, the VPC Resolver, private zones, hybrid endpoints, DNS Firewall, DNSSEC, query logs and Profiles, added one at a time until a company’s whole DNS fits on one map.
Kubernetes: how a cluster is put together
Tags:Other
The API server, etcd, controllers, scheduler, kubelet, Services, CoreDNS, volumes and node heartbeats, one part at a time, on a three-node cluster running one small app, with KCNA exam cues.
AWS access: how one laptop reaches a private database
Tags:AWS
Access keys, SigV4, STS roles, the credential chain, IAM Identity Center, Session Manager and RDS IAM auth, one part at a time, on the path from one laptop to a private database.
AWS KMS: how encryption at rest is put together
Tags:AWS
KMS keys, data keys, key policies, grants, CloudTrail, cross-account and multi-Region keys, rotation, key stores and deletion, added one at a time until the whole setup fits on one map.
AWS VPC: how a private network is put together
Tags:AWS
Subnets, route tables, security groups, network ACLs, NAT, endpoints, PrivateLink and Transit Gateway, added one at a time until the whole network fits on one map.
AI benchmarks: how a score is put together
Tags:AI
The task set, the grader, the harness, the score and its error bar, the runner, the leaderboard, the arena and the guards against leaked questions, one part at a time, from one coding agent’s run on SWE-bench Verified.
The AI stack: how an LLM application is put together
Tags:AI
The model call, LangChain, retrieval, LangGraph, the Claude Agent SDK, Langfuse and Airflow, one part per chapter, on one insurance-claims system.
Diffusion models: how noise becomes a picture
Tags:AI
The latent and its autoencoder, the seed, the denoiser that predicts noise and how it was trained, the sampler, the text encoder and cross-attention, guidance, and what newer models replaced, one part at a time.
MCP: how an AI app talks to its tools
Tags:AI
The host and its clients, the stdio pipe and JSON-RPC, versions in _meta, tools, resources and prompts, the call and its approval, the context window, elicitation, Streamable HTTP, OAuth and the host’s guardrails, one part at a time, on the 2026-07-28 revision.
Claude Code: how a coding agent is put together
Tags:AI
The parts of a coding agent, one per chapter: the loop around a stateless model, the request it sends, what loads before you type, the search tools, the prompt cache, the edit and its permission gate, hooks, subagents and compaction.
Python for LeetCode: the cheat sheet
Tags:Python
Copyable code blocks for loops and conditions, every operator including * and ** unpacking, the primitive types, the built-in functions, int and bit tricks, list, str, tuple, dict, set, Counter, defaultdict, deque, heapq, sorting and bisect, itertools, functools and math, plus ListNode, TreeNode, BFS, Dijkstra, union-find and trie templates, type hints and mypy, each type with what its operations cost.
Git worktrees: how three AI agents share one repo
Tags:AI
The checkout, the object store, refs, linked worktrees and the agent sessions on top of them, one part per chapter, until three agents editing one repository fit on one map.
Interpreted vs compiled: how a language runtime is put together
Tags:Python
The parser, bytecode, the interpreter loop, the profile, JIT tiers and deoptimization, the ahead-of-time compiler, the native binary and saved profiles, added one part at a time with one function in Python, TypeScript, Java and Go.
Reasoning models: how thinking is trained, run and paid for
Tags:AI
How a reasoning model is put together, part by part: the decode loop and the thinking block, the 0-or-1 checker and the RL loop that trained the habit, then the effort setting, the vote, the bill and the small students that copy it.
Concurrency: how a server holds 10,000 clients at once
Tags:Python
Ten thousand clients hit one 4-core server, and each waits 100 ms on a database. The parts that hold those waits, one at a time: sockets, threads, pools, epoll, event loops, runtime schedulers and Python’s lock, measured in Java, Node, Go and Python.
Transformers: how a language model is put together
Tags:AI
Every part of a transformer language model, one chapter each, on one map of GPT-2 small: tokenizer, embeddings, attention, MLP, output head, sampler, KV cache, and the encoder side of the 2017 original.
LoRA: how low-rank adapters fine-tune a frozen model
Tags:AI
How LoRA fine-tunes a frozen model, part by part: two thin matrices beside each weight, the side path they run on, which matrices get them, training, rank, and how the result is merged or served.
Prompt injection: how an AI assistant is hijacked and contained
Tags:AI
The parts of an AI assistant a prompt injection passes through, one per chapter (the model, the prompt, the inbox, retrieval and memory, tools, the exits), then the filter, tool scope, egress allow-list, quarantined reader and approval gate that contain it.
AI evals: how an eval pipeline is put together
Tags:AI
Traces, a golden set, a runner, code checks, human review, an LLM judge, statistics, a merge gate and a canary, one part per chapter, on one refund agent’s prompt change.
AI numbers every AI engineer should know: a model server, part by part
Tags:AI
The numbers that set what a model costs and how fast it answers, one part of the server at a time.
AI agent patterns: how a support agent is put together
Tags:AI
The model call, the request, the loop, tools and MCP servers, the router, memory, the plan, the policy gate, the approval queue, the output check and the trace store, one part per chapter, on one support agent refunding a double charge.
RAG: how a retrieval pipeline is put together
Tags:AI
A federal HR assistant answers from 2 million words of US law, part by part: the loader, chunker, embedding model, vector index, reranker and prompt builder that turn an $800 guess into a cited $1,500.
Neural networks: how a digit reader is put together
Tags:AI
The parts of a neural network, one per chapter, built around one handwritten 7: pixels, neurons, layers, softmax, loss, backpropagation, convolution and recurrence, with every number from one real run.
Loading a web page: how browser, network and server fit together
Tags:Other
Typing google.com into a fresh Chrome, part by part: the input path, the address bar, DNS, TCP, TLS 1.3 with a post-quantum key share, Google’s edge, HTTP/2, the servers behind it, the renderer, and what QUIC and the HTTP cache change.