25 July 2026
Sri Aurobindo’s ideas and the Mother’s practices decoded into plain English — 55 concepts, 10 books, 8 practices and a 54-term glossary, cross-linked by concept, book and practice with short cited quotations. Not self-help and not a shortcut; where scholars and practitioners disagree, it says so.
25 July 2026
A 22-chapter explainer of MCP — what the protocol actually says on the wire, the handshake, how authorization works, what MCP Apps are, the attacks specific to MCP, and the nine design patterns you reach for when putting an agent into production. Interactive diagrams throughout.
06 July 2026
An interactive essay on what makes Lisp different — code as data, the read-eval-print loop, and macros that let a program rewrite the language it is written in. Why those parentheses buy you a programmable programming language.
04 July 2026
An interactive, 13-chapter tour of the math that makes games work — vectors, Bézier curves, matrices, how GPUs turn math into pixels, flocking (boids), inverse kinematics, trig, particles, springs, shaders, 3D worlds, and a real-time-strategy build. Every idea is a live, playable demo.
02 July 2026
The tools got faster; the thinking didn’t get optional. Twelve interactive chapters on the parts of building software that AI can’t do for you — why you’re building, requirements and interviews, scope, writing it down, design, validation, edge cases, security, debugging, and working with AI.
02 July 2026
A visual course in one page on measuring stochastic software: datasets, graders, LLM-as-judge, the statistics that keep you honest, agent trajectories, regression gates, and the flywheel that turns failures into test cases. Fourteen chapters, every example drawn from one product.
01 July 2026
A chapter-by-chapter walk through the Maha Mrityunjaya Mantra (Ṛgveda 7.59.12): word by word, the ripe-gourd metaphor at its heart, the hidden negation (free me from death — but not from the deathless), sandhi and metre, and a chant-along sound engine.
26 June 2026
An interactive atlas of 151 conflicts from 1000–2024 CE on a scrubbable timeline, each drawn attacker→target — medieval invasions to modern wars, with top aggressors and casualty figures.
26 June 2026
How NASA and IBM built Prithvi-EO-2.0 — a 600-million-parameter foundation model that learns the planet from raw satellite imagery. Masked pretraining on Earth-observation data, and how one model transfers to floods, fires, crops and more.
26 June 2026
A visual, scrollable deep-dive into adapting a large model to your own needs — from the gradient that nudges a single weight, through LoRA and QLoRA, to preference tuning (RLHF and DPO). Each method, shown changing what the model learns.
23 June 2026
An interactive journey from a single NAND gate to a working computer that runs Tetris — logic gates and universality, XOR, binary arithmetic, memory, the CPU, then software, an assembler, and the game itself. Build the whole stack from the ground up, one layer at a time, in a day.
23 June 2026
A visual, math-from-scratch field guide to image processing — every filter, blur, and edge detector is just arithmetic on a grid of numbers. Convolution and kernels, blurs, gradients and edge detection, rebuilt from the ground up with no math past 10th grade, everything touchable.
23 June 2026
An interactive companion to how code actually runs — the full journey from source text to a running program, built around the Lox language from "Crafting Interpreters". Scanning, parsing and Pratt parsing, tree-walk evaluation, then a bytecode compiler and virtual machine — step through each stage transforming the code.
20 June 2026
An animated walk through the 2017 Transformer paper — why recurrence was the bottleneck, and how self-attention lets every word look at every other word at once, in parallel. Scaled dot-product and multi-head attention, sinusoidal positional encodings, the encoder–decoder stack, and the results that changed the field.
20 June 2026
A visual field guide to how modern language models are built — dense vs Mixture-of-Experts (active vs total parameters), decoder-only stacks, the KV cache and attention variants (MHA → GQA → MQA → MLA), RoPE, RMSNorm and SwiGLU — with real specs from GPT, Llama, Gemma, Qwen and DeepSeek.
20 June 2026
A look inside the black box: what actually happens to your prompt once it enters a language model. Tokens to embeddings, attention reading the whole context at once, the next-token decision — and how system prompts and rules steer every token, one prediction at a time.
20 June 2026
A visual, technical field guide to the GPU — from a single transistor to the silicon that trains frontier models. SIMT warps and massive parallelism, the memory hierarchy and HBM bandwidth, Tensor Cores and low-precision math (FP16/FP8/FP4), and why matrix multiply is the whole game.
20 June 2026
A field map of AI agents trying to become scientists — from literature-grounded research assistants to end-to-end autonomous discovery pipelines. Hypothesis generation, tool use and code execution, the reliability gap, and what the literature (PaperQA, The AI Scientist, Agent Laboratory and more) actually shows.
20 June 2026
The story of two protocols for wiring AI agents to tools. Anthropic’s Model Context Protocol — Host, Client, Server; tools, resources and prompts over JSON-RPC — and SCP, the Science Context Protocol from Shanghai AI Lab that scales the idea to a global web of scientific tools.
16 June 2026
A six-minute animated explainer on RAG — how you give a frozen language model fresh, private knowledge. From the problem (knowledge cutoff, no private data, fluent guessing) through embeddings, vector stores, retrieval with the same embedding model, chunking, and the actual augmented prompt — retrieve, then generate.
16 June 2026
A two-minute animated explainer video on Surya — NASA & IBM’s 366-million-parameter open foundation model for the Sun. From a solar storm racing toward Earth to the architecture that forecasts it: SDO’s 13 channels, 65,536-patch tokenization, long–short attention, a learnable Fourier filter, LoRA fine-tuning, and rollout forecasting — one core, many tasks, earlier warnings.
14 June 2026
An illustrated life of Sayagyi U Ba Khin (1899–1971) — Burma’s first Accountant General, who awakened amid his duties and taught the body-sweep vipassana of sensation, anicca, and equanimity. The disciplined accountant with the heart of a sage, whose student S. N. Goenka carried the method to the world.
14 June 2026
An illustrated life of Webu Sayadaw (1896–1977), the Burmese forest master who taught only one thing — awareness of the breath where it touches at the nostrils, in the midst of any activity. “Work, and be mindful.” His challenge, his eight-point practice, and the 1941 meeting with Sayagyi U Ba Khin that seeded the global vipassana movement.
14 June 2026
An illustrated life of Dipa Ma (1911–1989) — the householder master who awakened through devastating loss, in the midst of ordinary life: “Being a wife, being a mother — these were my first teachers.” Her story, her teaching of unbroken awareness, and her warrior principles.
14 June 2026
A slow, animated reading of fifty abhangas of Tukaram, the 17th-century Marathi saint-poet — the plain Marathi, romanization, the meaning underneath, and what to do with it on an ordinary day, framed by his life (the gāthā that surfaced from the river, the king’s gifts returned, Tukaram Beej).
13 June 2026
The mathematics behind machine learning, rebuilt from intuition — trigonometry, calculus, differential equations, vectors, matrices, and probability, culminating in building an LLM from scratch, with hands-on labs (a tiny transformer, a live bigram model, induction heads, and mechanistic interpretability). Every idea built by hand before the symbols arrive.
13 June 2026
A visual field guide to how Claude Code actually works — the agent loop, how the model itself decides to call a tool (stop_reason: tool_use), the transcript-as-memory and the context window, skills, the verify-and-retry loop, and sub-agents with their own context.
13 June 2026
A visual guide to how LLMs write — and when to trust them. Next-token prediction, the RLHF “house style,” the “I verified it” bluff, sycophancy and snowballing errors, a tiered catalog of AI-writing tells (delve, the em-dash, “it is worth noting”), and how to edit a draft back into a human voice.
13 June 2026
A contemplative field guide to the mind — the quiet, unbroken thread of awareness, built moment to moment. Twelve scenes on the momentum of continuous practice, drawn from the Dhammapada, the Yoga Sūtras, the Gītā, and teachers like Thich Nhat Hanh and Pema Chödrön.
13 June 2026
Say it, then say it better. An interactive guide to English phrasing for ESL writers — 64 patterns across the mechanics of tone, grammar foundations, sentence craft, professional email, Nonviolent Communication, and academic writing, each reshaping a plain draft into a sharper phrase.
13 June 2026
How a satellite turns sunlight into a number — and how you learn to trust it. A scrollable Earth-observation field guide: the measurement chain, sensors and orbits, then the statistics of defending a trend — Theil–Sen, Mann–Kendall, deseasonalizing, area-weighting, cross-validation, and when to refuse.
13 June 2026
A dozen interactive explainers turning the math and physics of Surya — NASA & IBM’s 366-million-parameter solar foundation model — into something you can touch: tokenization, attention, spectral gating, LoRA, rollout forecasting, skill scores, and the solar physics behind flares. Every widget runs real math.
13 June 2026
An interactive explainer on the craft of working with language models — context windows, prompting, the agent loop, compaction, retrieval, memory, tools, skills, and recursive models — in ten readable chapters.
13 June 2026
The same ideas as a scrubbable animated film — context windows, tokens, the loop, compaction, retrieval, tools and sub-agents, recursive models — that you can play at your own pace.
13 June 2026
What “vibe coding” actually is — building software by describing it and steering the model — as a short, scrubbable animated walkthrough.
Updated 13 June 2026
How do you check whether an AI is telling the truth? An animated walk through automated fact-checking — claim decomposition, retrieval, and the FactReasoner approach — with an interactive simulator, the math, and a video.
10 June 2026
A 7-minute animated walkthrough of what running an AI agent actually costs — tools, tokens, context, caching, RAG, and the monthly bill — in 19 chapters you can scrub like a movie.
13 June 2026
An animated exploration of emptiness and dependent arising, after the teachings of Rob Burbea — how perception is built, the “ways of looking” that shape what appears, fabrication and its dial, and how seeing this way loosens the grip of clinging.
12 June 2026
Sant Dnyaneshwar’s twenty-eight abhangas on a single radical idea — that simply saying God’s name is the whole of the spiritual path — with the poet, the Warkari tradition, and all twenty-eight verses in translation with word-by-word glosses.
26 July 2026
Four devotional recitation-texts glossed word by word — the Ganapati Atharvashirsha, the Bajrang Baan, the twelve names of Ganesha (Pranamya Shirasa Devam), and the Hanuman Chalisa: two for Ganesha, two for Hanuman. Every word glossed, every layer separable, with the claims of tradition and the findings of measurement (metre, matras, structure) set side by side and never dressed up as each other.