Light & Glow
Making a dark subject visible, one idea at a time.
Radiance for the age of intelligence
द्युति AI-Dyuti
dyuti · Sanskrit · radiance, light, splendor, brilliance
A library that switches the lights on across the whole arc of computing intelligence — from a first line of code to the last open question about minds.
The name
Why Dyuti?
Because most of what people call “artificial intelligence” is still a dark room. This is an attempt to find the switch.
Dyuti noun · Sanskrit · f.
Radiance, light, splendor, brilliance — in Sanskrit and many Indian languages; also a name for an inner glow.
Splendor & Majesty
Serious subjects deserve to be made beautiful.
An Inner Glow
Intelligence is best taught warmly, by someone who loves it.
तमसो मा ज्योतिर्गमय
“From darkness, lead me to light.”
Bṛhadāraṇyaka Upaniṣad · 1.3.28
Computing feels like magic where nobody turned the lamp on. AI-Dyuti turns it on.
The library
Two parts. One light.
Part I is the story and the intuition. Part II goes deeper and gets technical.
Part I Available now
RENOESIS
The Long Way Home
From a first line of code through data structures, algorithms, NP, the limits of computation, expert systems, machine learning, the transformer, quantum computing and agents — then back to natural intelligence, which we still do not understand.
30Chapters
8Parts
30Drawn figures
533Defined terms
Part II In preparation
The Technical Companion
Coming soon
The deeper, more technical half — picking up where Part I’s intuition ends.
Lighting the lamp…Inside Part I
Eight parts, one continuous argument
Written to be read in order — every chapter hands something to the next. Step in at any door.
I
The Craft
From typing code to architecting minds.
Ch 1–4
II
The Shape of Problems
Data structures, algorithms, and the wall called NP.
- 05The Sock DrawerData structures, and why shape is destiny
- 06How Fast Does the Pain GrowComplexity, Big-O, and the only maths that will save you
- 07Four Ways to ThinkDivide and conquer, greedy, dynamic programming, backtracking
- 08The WallP, NP, NP-complete, NP-hard, and the most important open question in science
Ch 5–8
III
The Limits
What computation is, and what no machine can ever do.
- 09Four CagesAutomata, languages, and the Chomsky hierarchy
- 10Where the Theory Earns Its RentCompiler design, from text to a running machine
- 11The Things No Machine Can DoTuring, the halting problem, undecidability, and Gödel's shadow
- 12Generate and TestThe oldest idea in artificial intelligence, and the newest
Ch 9–12
IV
The Old Gods
Facts, axioms, expert systems, and the winter that followed.
- 13A Fact, Written DownFacts, rules, axioms, hypotheses, and the machinery of inference
- 14Dr. Rao at Two in the MorningMYCIN, the expert system that worked and never ran
- 15Where the Dream BrokeMonotonicity, closed worlds, frames, and the brittleness of rules
- 16Letting Go of CertaintyFuzzy sets, probability, rough sets, evolution, and the first neurons
Ch 13–16
V
The Ladder of Knowing
From a scrap of paper to a weight in a neural network.
- 17The Ladder of KnowingFrom a scrap of paper to a weight in a neural network
- 18How a Pile of Numbers LearnsMachine learning and deep learning, from first principles
- 19One Token's JourneyThe transformer, explained all the way down
- 20Teaching a Liar to Cite Its SourcesRAG, fine-tuning, context, and the engineering of trust
Ch 17–20
VI
The Quantum Detour
What a qubit actually buys you, and what it does not.
Ch 21–22
VII
Agency and Institutions
Agents, and the architecture of hospitals, banks and states.
- 23The Thing That ActsAgents, tools, memory, and why every old architecture came back
- 24Dr. Rao Gets Her SystemA reference architecture for healthcare, checked against MYCIN's ghost
- 25The Machine That Must Explain ItselfFinance — risk, fraud, and explanation as law
- 26The Village ClerkPublic administration, where a wrong answer has no appeal
- 27Who WatchesGovernance, evaluation, and the institutions we have not built yet
Ch 23–27
VIII
The Return
Convergence, democratization, and the long way home.
Ch 28–30
Click a chapter to open it, in the language you chose.
A first taste
It begins in a kitchen
My mother cooks dal from memory, without measuring. Ask for the recipe and she will say there isn’t one. “You just know when it’s ready” — true for her, useless for me. It took forty minutes to write down what she does in fifteen: wash the lentils, pick out the stones, two fingers of water, lower the flame when the foam comes.
— Chapter 1, The Recipe and the Cook
How it is made
Four promises
Light that is honest about what it shows.
Nothing assumed
Every term is defined where it first appears — hover a dotted word, or open the Lexicon.
Drawn, not decorated
Thirty vector figures, sharp at any size, full-screen on click.
Light on the wire
No trackers, no external fonts, no network calls inside the book.
One argument
Chapters are handoffs. The margin links show how everything connects.