🧭 Start Here

Where should you begin?

This site has 120+ long-form articles across more than a dozen series. Pick the track that fits you and read in order — each step builds on the last, and every series article ends with a link to the next one.

Interested in the hardware, energy and money behind AI? 🖥️ See the AI Compute Stack at a glance →

Want to see how a frontier model is actually made? 🏭 See Producing Intelligence at a glance →

Curious who actually runs these machines? 🧑‍🔧 See Who Runs the AI Factory at a glance →

Want to compare the world's biggest AI factories? 🏗️ See AI Factories at a glance →

Want an AI factory of your own? 🧰 See AI Factory as a Service at a glance →

Curious about the software inside? 🧠 See Inside the ML Layer at a glance →

Where does training data come from? 🗃️ See Inside the Data Layer at a glance →

How are models tested before they ship? 🎯 See Evaluation at a glance → · What lets agents act safely? 🤖 See the Agent Layer at a glance →

See the whole machine at once? 🏭 Open The AI Factory, Cut Open → · 🛰️ From Above → · 🏗️ In 3D → · 🧬 Software stack, live → · 🧬 Stack Innovation Map →

🌱 Understand how it works

For curious readers, students and anyone switching in. About 6 hours of reading.

  1. 1
    Build a GPT from scratch, explained plainly
    The interactive course in Layman mode: what a language model actually is, how it learns, and how it's trained — no jargon.
    Interactive course · self-paced
  2. 20
    The complete stack (interactive)
    Everything from transformer to production agent on one animated map — follow a request through every layer.
    10 min
  3. 2
    How a transformer works, end to end
    Follow one sentence through a GPT-style model step by step, with interactive demos and code you can run — including how training works.
    45 min
  4. 3
    Reasoning models & test-time compute
    Why the newest models "think" before answering, and why that became the second scaling law.
    25 min
  5. 4
    Multimodality & the great consolidation
    How text, image, audio and video models are merging into one.
    26 min
  6. 5
    AI agents & the enterprise agent economy
    What changes when models stop answering questions and start doing tasks.
    25 min
  7. 6
    AI safety, alignment & interpretability
    The open problems that decide whether any of this goes well.
    25 min
  8. 7
    The GPT lineage, GPT-1 to GPT-6
    One model family's full history — the clearest way to see how the field actually progressed. Other labs' lineages follow via the Next links.
    85 min · Model Lineage series
  9. 8
    The research frontier map
    What every major lab is betting on right now, side by side.
    75 min

🔬 Build the skills to do the research

For students and engineers aiming at research or research-engineering roles. Weeks, not hours — work the exercises.

  1. 1
    Self-assessment: find your gaps first
    Check what you already know so you can skip it — then fill only the gaps.
    65 min · AGI Researcher Foundations
  2. 2
    The mathematical foundations
    The linear algebra, probability and optimization you'll actually use.
    75 min · AGI Researcher Foundations
  3. 3
    Core deep learning concepts
    Then continue through the Foundations curriculum — RL, interpretability, alignment, world models, systems — using the Next links.
    75 min · AGI Researcher Foundations
  4. 4
    Build a language model from scratch (Stanford CS336)
    A lecture-by-lecture companion with runnable code: tokenizer, resource accounting, architecture, and on through training.
    20 min to start · Language Modeling from Scratch series
  5. 5
    How pretraining works, end to end
    From raw web text to a foundation model: data filtering, deduplication, the training recipe and the systems that scale it — with a pipeline you can run.
    40 min
  6. 6
    How fine-tuning works: full fine-tuning vs LoRA
    Turning a base model into an assistant or specialist, with a runnable experiment comparing the two main methods.
    40 min
  7. 7
    How post-training works: SFT, RLHF, DPO and RLAIF
    How a base model becomes an aligned assistant, with an experiment comparing SFT, DPO and RLHF on the same model.
    45 min
  8. 8
    How inference works: serving a model at scale
    Prefill and decode, the KV cache, batching, quantization and speculative decoding, with a mini engine you can run.
    40 min
  9. 9
    Scaling laws, GPUs and distributed training
    Why bigger models work, what GPUs are fast at, and how one model is split across thousands of them, with a mini scaling study you can run.
    40 min
  10. 10
    Production LLM system design
    Everything around the model in a ChatGPT- or Claude-style assistant: context building, memory, tools, safety layers, caching and streaming, with a runtime you can run.
    35 min
  11. 11
    Context engineering and memory
    What goes into the context window and how it is chosen, compressed and remembered, with five memory strategies measured side by side.
    30 min
  12. 12
    RAG: retrieval-augmented generation
    How assistants find and cite the right documents — chunking, embeddings, hybrid search, reranking and evaluation, measured on real data.
    35 min
  13. 13
    Tool and function calling
    How models go from answering to acting: schemas, the call loop, permissions, retries and prompt-injection defence, with an orchestrator you can run.
    30 min
  14. 14
    Agentic systems
    The plan-act-observe loop, subagents, human approval and stopping rules, and why reliability decides how long a task an agent can finish.
    30 min
  15. 15
    Multi-agent systems
    When a team of agents beats one agent — and when it doesn't: roles, routing, shared state, parallelism and cost, measured.
    30 min
  16. 16
    Agent memory and planning patterns
    How long-running agents keep state, survive crashes, learn from experience and replan, with each effect measured.
    30 min
  17. 17
    MCP and tool ecosystems
    The Model Context Protocol on the wire: how agents discover and use tools, resources and prompts, with a server you can run.
    30 min
  18. 18
    Production agent architecture
    How everything comes together into a reliable platform: control and execution planes, durable execution, routing, tracing and scaling.
    30 min
  19. 19
    Evaluating agents
    How to tell whether an agent really works: trajectories, environment checks, judges, pass^k reliability and release gates.
    30 min
  20. 20
    Harness engineering
    Why the system around a model decides what it can do — measured with one model inside five harnesses.
    20 min
  21. 21
    How to read a paper like a researcher
    A repeatable method for getting through the literature without drowning in it.
    60 min
  22. 22
    Day one at a frontier lab
    The engineering practicum: production training code, distributed debugging, kernels, on-call.
    55 min · Frontier Lab Engineering Practicum
  23. 23
    Recursive self-improvement: a primer
    One live research frontier, end to end — then the CS329A self-improving agents handbook.
    45 min · RSI series
  24. 24
    The researcher's field guide: getting hired
    Papers, labs, people and the hiring process — plus the study stack and PhD programs guides.
    80 min

🚀 Find the opportunity and build in it

For founders, operators and investors. About 6 hours, then the 12-part startup series.

  1. 1
    The six real levers of frontier AI
    The framework the rest of this track uses: what actually moves capability forward.
    25 min · AGI Capability Stack series
  2. 2
    The AGI capability stack, mapped
    Where today's frontier models stand on each layer, and what closing each gap takes.
    28 min
  3. 3
    The levers nobody's pulled yet
    Where a new lab could realistically win against incumbents.
    30 min
  4. 4
    Compute frontier: why clusters keep growing
    The cost structure underneath everything: training runs, mega-clusters, hardware.
    24 min
  5. 5
    Frontier model pricing
    What the models cost to use, and what that means for product margins.
    20 min
  6. 6
    AGI startups to watch
    Who's already building in the open lanes.
    28 min
  7. 7
    The three startup lanes
    Frontier scale-first, the applied & agentic layer, and the narrow research bet — strategy and team for each.
    50 min · Lane Deep Dives series
  8. 8
    Founding an AGI startup: the 12-part series
    From finding the idea through legal, team, seed, product, go-to-market and Series A.
    12 parts · Founding an AGI Startup series
Prefer to browse? Every series is listed on the homepage, and All Posts has the full archive. Your ticked steps are saved in this browser only.