Field guides
Notes from the systems side.
I spent my first career as a commercial trial lawyer. Four years ago I joined Amazon on the security side — first writing information-security policy for customer-service operations, then as a security engineer building GenAI tooling to scale third-party security assessments. Today I'm an AI product manager for that same business's trust-and-compliance team, building tools that make regulatory compliance — the EU AI Act, privacy, accessibility, data governance — scalable and intelligent across operations spanning more than twenty countries: one of the largest, if not the largest, customer-service organizations in the world.
These 33 courses are the field guides I needed along the way. Each opens with a diagnostic that reorders the syllabus around what you don't know yet, then works through modules with knowledge checks, applied worksheets, and a spaced-review queue. Progress is saved in your browser — and shows up right here on each card — so nothing leaves your machine.
New here? Three ways in.
Each path is an ordered reading list — take it left to right.Building with LLMsfrom model to measurable product
Security for agentic systemswhat can go wrong, and who may do what
Ship with engineers, as a PMspeak the room's language
AI & Language Models
Building on models, and proving they work. 07Building on Models
Construct, spend, and defend the context budget every LLM feature comes down to
8 modulesStart → Nº03Proof of Quality
Golden sets, calibrated judges, and CI gates that prove an LLM system works
8 modulesStart → Nº08Trust, then Verify
Reviewing code you didn't write — and when a coding agent earns autonomy
8 modulesStart → Nº28Specs as the New Source Code
Writing requirements an AI can execute
9 modulesStart → Nº11The Meter Is Running
Unit economics of AI products
8 modulesStart → Nº13Under the Hood
Machine learning beyond the chat window
8 modulesStart → Nº04From Statute to Spec
Operationalizing AI regulation as product requirements
8 modulesStart →Security & Identity
Who may do what — and how it goes wrong. 08Hostile Inputs
Threat modeling agentic AI
9 modulesStart → Nº29The Confused Deputy Comes for AI
Why agents can't tell instructions from data — and how to bound the blast radius
8 modulesStart → Nº05Deputies and Delegates
Authorization in the age of agents
8 modulesStart → Nº15Who Goes There
Authentication and identity
8 modulesStart → Nº30The Unphishable Login
Why a hardware security key can't be phished
8 modulesStart → Nº19Locks and Keys
What "encrypted" actually means — and when it's a lie
8 modulesStart → Nº23How Systems Get Hacked
Thinking in boundaries
8 modulesStart → Nº27Borrowed Code
Open source & the dependency supply chain
9 modulesStart →Systems & Architecture
Shapes that hold under load. 05Serverless Systems
Serverless design on AWS
8 modulesStart → Nº24Drawing Boundaries
Coupling, boundaries, and what a service split actually buys — and costs
8 modulesStart → Nº17Later, Not Never
Queues, events, and asynchronous systems
8 modulesStart → Nº09Systems That Confess
Three signals, SLOs, error budgets — and alerting that respects the pager
8 modulesStart → Nº31Fast Enough
Making systems fast — honestly
8 modulesStart →Data & Analytics
Storing it, querying it, and believing the answer. 04The Shape of Data
Schemas as enforced invariants — normalization, indexes, transactions, real SQL
8 modulesStart → Nº10Beyond the Dashboard
Causal traps, statistical power, and honest reads of ambiguous results
9 modulesStart → Nº20Where the Data Goes Next
From production data to a dashboard number you can actually trust
8 modulesStart → Nº22Finding Things
Search, from keywords to vectors
8 modulesStart →How the Web Works
The machinery under the page. 04The Life of a Request
HTTP and the network beneath it
9 modulesStart → Nº21After the Bytes Arrive
Rendering, the event loop, origins — and reading it all in devtools
8 modulesStart → Nº16Where Code Runs
Servers, containers, and the cloud
9 modulesStart → Nº25What Code Is
Machinery under your source code
8 modulesStart →Building & Shipping
From contract to commit to customer. 03Contracts at the Boundary
Errors, pagination, idempotency, versioning — the clauses integrators judge you by
8 modulesStart → Nº18From Commit to Customer
Version control and CI/CD
8 modulesStart → Nº26What Green Means
What a green suite actually claims, where it lies, and what to test first
8 modulesStart →Product
Judgment about what deserves to exist. 02Why these exist. The thread through every role — courtroom, security, compliance, product — is the same: pattern matching and systems thinking, high judgment applied to evidence at scale. A trial applies it to a record. Compliance applies it to controls and regulations. Product applies it to customer problems. GenAI finally makes that judgment operational instead of artisanal, and these courses began as my own study notes while learning to build that way — which is why the worked examples keep reaching for the first career: cross-examination, chain of custody, pleading standards. I rewrote them as the courses I wish I'd found — honest about what's hard, and testable, so you can't fool yourself — because the surest test of a mental model is whether you can teach it.
Your progress
Everything is saved in this browser's local storage — no server, no account. Export writes a small JSON file of your course progress; import it in another browser to carry the same state over. Your theme choice is a device preference and isn't included.