Experienced leaders
AI for engineering leaders
Engineering strategy, AI-native delivery, team design, evaluation, and governance, in the order I would read them.
Not startedStart reading
Study notes
When I want to learn a topic properly, I put together study material on it with AI tools. These are those notes, mostly on AI and engineering, shared in case they help someone else.
These are personal notes, not Coursera courses, and nobody has reviewed them. For courses taught by universities and companies, go to Coursera.
Suggested orders
If you want an order to read in, these group related notes. Progress is saved on this device.
Advanced builders
Building production AI
From agent architecture and tools to evaluation, observability, security, and durable execution.
Not startedStart reading
Foundational to intermediate
AI foundations
How models work, how they become assistants, how progress is measured, and how governance is evolving.
Not startedStart reading
Start here
Just added
All notes
/
44 notes
How models work
The Mechanism — What a Transformer Actually Computes
Part 1 of a five-part series on how LLMs work. The forward pass step by step: attention, the residual stream, and where the parameters live.
43 min read · Updated Oct 17, 2026 New
The Training Stack — What It Costs to Make a Model
Part 2. Scaling laws worked out as unit economics: training compute, memory per parameter, parallelism, and why Chinchilla's headline fit didn't replicate.
40 min read · Updated Oct 17, 2026 New
Post-Training — How a Document Completer Becomes an Assistant
Part 3. SFT, RLHF and DPO, with DPO's closed form derived step by step, and what that shortcut costs.
36 min read · Aug 28, 2026 New
Evals — Error Analysis First, and Judges You Have Validated
Part 4. Read traces and label the failures before picking a metric, then check an LLM judge like any other classifier before trusting its pass rate.
47 min read · Updated Sep 30, 2026 New
Inference Economics — Why Decode Is Bandwidth-Bound and What It Costs
Part 5. Why decoding is limited by memory bandwidth, why the KV cache caps how many users a server can hold, and what that does to serving costs.
37 min read · Aug 28, 2026 New
Building with AI
The New Rules of Context Engineering for Claude 5 Models
Based on Thariq's July 2026 article on how prompting changes for Claude 5 models, with applied modules on CLAUDE.md, skills and tool design.
71 min read · Jul 26, 2026 New
The Context Budget — Tokens, Skills and Memory as an Enterprise Decision
Skills, CLAUDE.md, memory, retrieval, MCP and subagents compared as different ways to decide what the model sees, and what each one costs per turn.
60 min read · Updated Sep 15, 2026 New
Claude Mods — Reading a Feature Before It Ships
How Claude Code's extension model is moving from hooks that vote on an action to mods that wrap it, read from a public issue and published source before release.
32 min read · Updated Oct 2, 2026 New
What to Do About Slop — The Cherny Exchange, Examined
A close reading of the email exchange Boris Cherny published about AI-written code in an old codebase, and what each side actually argued.
30 min read · Updated Sep 15, 2026 New
The Attention Factory — Reading OpenAI's Agentic Pipeline
OpenAI's nine-stage agent pipeline, read as a system for deciding what a person should look at rather than a system for writing code.
47 min read · Sep 15, 2026 New
The AI-Native SDLC, Audited
Anthropic's playbook for running each stage of software delivery with agents, checked against METR, DORA 2025 and other published evidence.
40 min read · Updated Oct 2, 2026 New
Choosing and Tuning Claude 5.5 Models
When Sonnet 5.5 is the cheaper pick and when it isn't, what a task really costs on Opus 5.5, and how effort settings change the bill.
18 min read · Updated Oct 14, 2026 New
Designing Agents Around the Cache, the Tools and the Skills
Lessons from the Claude Code team on prompt caching, tool design and skills, with the cache treated as part of the architecture and priced out.
19 min read · Updated Oct 3, 2026 New
Harnesses That Fit the Task — Dynamic Workflows and Measured Agent Loops
Six harness patterns for agents, quarantining untrusted input, and how to run a measured improvement loop with benchmarks, flags and staged rollout.
13 min read · Oct 1, 2026 New
AI Evals and Measurement
Eval design, datasets, LLM-as-judge, offline and online testing, and how to read results without fooling yourself.
35 min read · Jun 30, 2026 New
Coding Agents — Using AI to Build Software
The 2026 coding agents side by side (Claude Code, Cursor, Copilot, Kiro, Devin and others), with prompting, day-to-day workflows and team practices.
168 min read · Updated Oct 14, 2026 New
Reviewing Agent-Written Code — Risk Tiers, AI Reviewers and Convergence
How the Claude Code team reviews code, read against the product docs: approval by risk tier, AI reviewers, and what to do when review never converges.
19 min read · Oct 16, 2026 New
Claude Certified Architect — Exam Prep
The CCA-F exam's five domains, with study notes, practice questions and a study timeline.
113 min read · Jul 12, 2026 New
Agents in practice
Loop Engineering — Designing Systems That Run AI Agents
Self-correcting loops for coding agents, drawn from Addy Osmani, Boris Cherny, Geoffrey Huntley and Peter Steinberger, with code examples and short animations.
88 min read · Updated Jul 18, 2026 New
Scheduled Agents That Fail Loudly
Building a daily agent automation that neither repeats nor drops work: bookmarks, ordering of side effects, credentials and budget caps.
13 min read · Oct 9, 2026 New
Delegating to Cloud Agents — Proof, Parallelism and the Trust Boundary
What cloud agent sessions are good for, what parallel runs cost in merges and review time, and which delegated workflows carry the most risk.
16 min read · Oct 7, 2026 New
Durable Execution for Agents — Temporal, Vercel Workflows, Inngest, Restate and DBOS
Five durable-execution platforms compared for agent workloads from their own docs and pricing, with crash recovery, approval gates and spending caps.
99 min read · Oct 14, 2026 New
Computer-Use and Browser Agents — Screens, Sandboxes and the Permission Problem
Pixels versus accessibility trees, cost per task, how to read the benchmarks, sandboxes, sign-in, prompt injection, and build versus buy.
65 min read · Oct 13, 2026 New
Claude Agent SDK
A hands-on guide to building agents with the Claude Agent SDK: tool use, multi-turn conversations and orchestration patterns.
43 min read · Updated Oct 31, 2026 New
MCP and the Agentic Tooling Stack
The Model Context Protocol and what sits around it: primitives, transports, tool design, building servers, multi-agent orchestration and security.
32 min read · Jun 30, 2026 New
Agent Tracing and Observability
Traces, spans and metrics for agent systems, and how to debug and tune them in production.
46 min read · Updated Oct 14, 2026 New
AI Agents in Production
Agent architectures, tools, memory, guardrails, evaluation, cost and observability, with case studies of agents running in production.
73 min read · Jun 29, 2026 New
OpenAI Realtime API
Building low-latency voice and text apps with the Realtime API: streaming, WebSockets and function calling.
14 min read · Jun 13, 2026 New
Security and compliance
Securing Agents — Prompt Injection, Sandboxes, Egress and Credentials
The lethal trifecta, prompt-injection defences that hold up, least-privilege tools, sandbox egress, agent credentials and kill switches.
71 min read · Oct 13, 2026 New
The AI Compliance Map for a US SaaS Company
Which AI laws bind a US SaaS company with EU users and US enterprise customers as of October 2026: the EU AI Act, the federal preemption fight, Colorado, California and Texas.
100 min read · Oct 14, 2026 New
AI progress
Price and Pace — Measuring AI Progress from the Outside and the Inside
Epoch AI's estimate of how fast the cost of AI is falling, next to Anthropic's proposal for measuring progress from inside a lab.
20 min read · Sep 26, 2026 New
We Must Pace the Frontier — Amodei's Argument, Examined
Dario Amodei's September 2026 essay laid out as a chain of claims, then tested against its sources and the main objections.
67 min read · Updated Sep 15, 2026 New
Jev — AI That Doesn't Talk
TypeSafe AI's Jev returns typed answers with probabilities instead of chat. What its calibration does and doesn't guarantee, and the economic bet behind it.
19 min read · Sep 27, 2026 New
Suncatcher — The Engineering Case for AI Compute in Orbit
The engineering in Google's paper on putting TPUs in orbit: power, laser links between satellites, and radiation, with measured results kept apart from projections.
15 min read · Sep 26, 2026 New
The 2026 AI Model Landscape
The model families as of July 2026: mixture-of-experts, open and closed weights, reasoning, multimodal and small models, and what they cost.
127 min read · Updated Oct 20, 2026 New
AI Governance — Hassabis's FINRA for AI
Demis Hassabis's July 2026 call for a self-regulatory body for AI, set against Amodei's and Altman's proposals, with the history behind the debate.
40 min read · Jul 19, 2026 New
AI Frontier Voices — Learning from the People Building It
Fifteen people at Anthropic, OpenAI, Google, Cursor and xAI worth following on X, what each one works on and what they've been saying.
36 min read · Updated Nov 2, 2026 New
Engineering leadership
The Agent-Era Engineering Org — Team Shape, Ladders and Reviews
Reorganising engineering when agents write most of the code: the unit of work, reviewer capacity, junior roles and ladders, hiring, metrics, and a 90-day plan.
57 min read · Oct 13, 2026 New
The Forward Deployed Engineer Wave
Seven organisations launched forward-deployed engineering teams in 2026. A practitioner's argument that they'll fail the same way, fact-checked claim by claim.
45 min read · Aug 28, 2026 New
Engineering Leadership at Scale
Managing managers, org design, hiring, performance, culture, planning, reorgs, remote teams and developer experience.
73 min read · Jun 29, 2026 New
Platform Engineering
Internal developer platforms, CI/CD, Terraform and Kubernetes, OpenTelemetry, SRE practice, feature flags and database migrations.
67 min read · Jun 29, 2026 New
Systems and infrastructure
System Design at Scale
Distributed systems, sharding, caching, load balancing, queues and the CAP theorem, worked through five classic design problems.
74 min read · Jun 29, 2026 New
Data Engineering and Analytics for Leaders
The modern data stack: warehouses, dbt, Airflow, Kafka and Flink, data mesh versus lakehouse, experimentation infrastructure and feature stores.
87 min read · Jun 29, 2026 New
Cloudflare for Developers — The Full-Stack Edge Platform
Workers, KV, R2, D1, Pages, Workers AI and Durable Objects, and why coding agents make Cloudflare a good default for internal tools.
69 min read · Jun 30, 2026 New
Nothing matches. Try another topic or length.