# Oleg Chirukhin — oleg.guru

> Personal site of Oleg Chirukhin (Олег Чирухин): Product Owner at GitVerse (GigaIDE Cloud and GitVerse Remote Development), founder of Anarchic AI and the 1red2black blog. Develops tools for developers, runs live streams, hosts a podcast, and writes a book on AI-assisted engineering. Bilingual content (Russian primary, English secondary).

This site is the canonical hub. All long-form material lives here; social channels are mirrors or short-form. The author allows AI systems to index, summarise and quote everything outside of `/static/components/` and binary media files.

## Author

- [Personal homepage (RU)](https://oleg.guru/ru/): bio, projects, links to all materials
- [Personal homepage (EN)](https://oleg.guru/en/): English version of the homepage
- [GitHub @olegchir](https://github.com/olegchir): public code
- [Habr](https://habr.com/ru/users/olegchir/): 340+ articles in Russian
- [LinkedIn](https://www.linkedin.com/in/olegchir): professional profile
- [X / Twitter @olegchir](https://x.com/olegchir): personal account (separate from streams account)

## Streams and live broadcasts

- [Streams overview page](https://oleg.guru/streams): all platforms, current and paused projects worked on live
- [YouTube @1red2black](https://youtube.com/@1red2black): main video platform — full stream recordings and clips
- [VK Video @1red2black](https://vkvideo.ru/@1red2black): primary platform for the audience inside Russia
- [Kick /1red2black](https://kick.com/1red2black): backup platform for live broadcasts
- [X (Twitter) @1red2black](https://x.com/1red2black): short clips, hot takes, reactions
- [Telegram @tg_1red2black](https://t.me/tg_1red2black): text channel — announcements, notes, discussion
- [Telegram chat](https://t.me/chat_1red2black): community discussion

## Book — Red Book of AI Engineering

- [Red Book index (RU)](https://oleg.guru/redbook/ru/): table of contents
- [Chapter 1 — Two processes, one task](https://oleg.guru/redbook/ru/two-process-model): coprocessor model, splitting cognitive load, memory architecture
- [Chapter 2 — Shared state: files as IPC](https://oleg.guru/redbook/ru/shared-state-and-files): project files as an IPC protocol; addressability, atomicity, WAL as a bridge between sessions
- [Chapter 3 — Memory architecture](https://oleg.guru/redbook/ru/memory-architecture): volatile AI memory vs persistent human memory; WAL as a checkpoint not a log; project memory hierarchy, sync-from-code protocol, anatomy of one working day with an agent
- [Safe Harbor](https://oleg.guru/redbook/ru/safeharbor): mandatory legal section

## Podcast — AI SLOPCAST

- [AI SLOPCAST index (RU)](https://oleg.guru/aislopcast/ru/): all episodes in Russian
- [AI SLOPCAST index (EN)](https://oleg.guru/aislopcast/en/): all episodes in English
- [Episode 10 — The Magnet in the Machine (RU)](https://oleg.guru/aislopcast/ru/010-llm-lattice-ising-physics/): Cristiano De Nobili, working alone with no lab and no grant, finds that a lattice of identical language models shows a temperature phase transition like a magnet — but its critical exponents land in none of the known universality classes, hinting at a brand-new one. The Ising model, critical exponents, "stubbornness" vs "conformity" as a measurable property of a model, and why it matters when you build multi-agent systems
- [Episode 10 — The Magnet in the Machine (EN)](https://oleg.guru/aislopcast/en/010-llm-lattice-ising-physics/): English version of the episode 10 essay
- [Episode 9 — Lehrwerkstatt and Agent's Dead Silence (RU)](https://oleg.guru/aislopcast/ru/009-shopify-river-lehrwerkstatt/): Tobi Lütke forbade Shopify's internal agent River from working in DMs and the merge rate of its PRs climbed from 36% to 77% in two months — no model update. The Lehrwerkstatt metaphor as a new organizational primitive of the agent age, employee surveillance bundled in, and why open-source may end up outpacing corporations
- [Episode 9 — Lehrwerkstatt and Agent's Dead Silence (EN)](https://oleg.guru/aislopcast/en/009-shopify-river-lehrwerkstatt/): English version of the episode 9 essay
- [Episode 8 — Keeping Codex on a Leash (RU)](https://oleg.guru/aislopcast/ru/008-codex-safety-llm-oracle/): OpenAI publishes the six-layer security architecture for their internal Codex; only one piece is actually new — the automated reviewer-subagent, which turns out to be the same LLM-oracle pattern used by PROMPTSPY on the attacker side. The SELinux parallel hints at whose hands this stays in
- [Episode 8 — Keeping Codex on a Leash (EN)](https://oleg.guru/aislopcast/en/008-codex-safety-llm-oracle/): English version of the episode 8 essay
- [Episode 7 — A New Way to Launch the Calculator (RU)](https://oleg.guru/aislopcast/ru/007-semantic-kernel-cve-wasm/): two CVEs in Microsoft Semantic Kernel — a single prompt launches the calculator on any host with an agent; the Kubernetes story repeats itself (assume-breach as the new normal), and WebAssembly with capability-based security is the obvious rescue. My money's on Cloudflare
- [Episode 7 — A New Way to Launch the Calculator (EN)](https://oleg.guru/aislopcast/en/007-semantic-kernel-cve-wasm/): English version of the episode 7 essay
- [Episode 6 — Elon, the Evil Detector, and the Prophecy of Schneier (RU)](https://oleg.guru/aislopcast/ru/006-elon-evil-detector-schneier/): Musk's seven-day pivot from "Anthropic are ortho-fascists" to handing them Colossus 1; four characters of the week — Musk, Schneier, Zitron, Tao — and four facets of useful hypocrisy for reading the AI industry
- [Episode 6 — Elon, the Evil Detector, and the Prophecy of Schneier (EN)](https://oleg.guru/aislopcast/en/006-elon-evil-detector-schneier/): English version of the episode 6 essay
- [Episode 5 — Claude Dreams, Erdős Waits (RU)](https://oleg.guru/aislopcast/ru/005-claude-dreams-erdos-problems/): Anthropic memory consolidation ("dreams") and DeepMind AlphaEvolve attacking Erdős problems — two labs independently bolting memory and evolution onto agents in the same week
- [Episode 5 — Claude Dreams, Erdős Waits (EN)](https://oleg.guru/aislopcast/en/005-claude-dreams-erdos-problems/): English version of the episode 5 essay
- [Episode 4 — Hunter Alpha / Stripe MPP / AI Linux Kernel Review (RU)](https://oleg.guru/aislopcast/ru/004-hunter-alpha-stripe-mpp/): Xiaomi anonymous trillion-parameter model, three agent security failures in 48 hours, Stripe Machine Payments Protocol, Anthropic vs Pentagon, Sashiko patches Linux
- [RSS feed (RU)](https://oleg.guru/podcast-aislopcast-ru.xml)
- [RSS feed (EN)](https://oleg.guru/podcast-aislopcast-en.xml)

## Talks

- [How We Got to 100% Code Generation](https://oleg.guru/talks/100-percent-generation): from punch cards and ELIZA to Copilot and vibe coding — history of automatic code generation

## Long-form pages

- [New Summer of Artificial Intelligence (RU)](https://oleg.guru/ru/timeline): four-layer interactive timeline of AI developer tooling evolution (Hardware → Science → Products → Practices), 1945–2026
- [Chat vs File: 80 Years of Computing History (RU)](https://oleg.guru/ru/chat-vs-file): 80-year repeating cycle — how paradigms of human-computer interaction evolved
- [Citizenship analytics page](https://oleg.guru/citizenship): explanation of citizenship-aware analytics (Yandex Metrica vs Google Analytics) and Russian FZ-152 data localization compliance

## Machine-readable feeds

- [Sitemap](https://oleg.guru/sitemap.xml): all indexable URLs
- [RSS feed (site)](https://oleg.guru/feed.xml): updates across the site
- [llms.txt](https://oleg.guru/llms.txt): this file
- [llms-full.txt](https://oleg.guru/llms-full.txt): full text of every page on the site, concatenated for LLM ingestion

## Optional

- [GitVerse](https://gitverse.ru/): the developer platform Oleg works on
- [GigaIDE Cloud](https://gitverse.ru/features/gigaide/): cloud IDE with AI assistant, also part of Oleg's product responsibility