<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>Yegor Gaidar — Writing</title>
    <link>https://lobsterswamp.space/blog.html</link>
    <description>Technical writing on distributed systems, AI infrastructure, data platforms and on-device AI — with the numbers, not the vibes.</description>
    <language>en</language>
    <managingEditor>maingaidar@gmail.com (Yegor Gaidar)</managingEditor>
    <webMaster>maingaidar@gmail.com (Yegor Gaidar)</webMaster>
    <lastBuildDate>Fri, 28 Aug 2026 00:00:00 +0000</lastBuildDate>
    <atom:link href="https://lobsterswamp.space/feed.xml" rel="self" type="application/rss+xml"/>
    <image>
      <url>https://lobsterswamp.space/img/og-card.png</url>
      <title>Yegor Gaidar — Writing</title>
      <link>https://lobsterswamp.space/blog.html</link>
    </image>

    <item>
      <title>I Trained a 1.7B Model to Beat an 8B One — Crab-1 OSINT Agent</title>
      <link>https://lobsterswamp.space/blog/crab1-slm-osint/</link>
      <description>How a distilled 1.7B SLM (Qwen3 + QLoRA SFT) outperformed Llama 3.1 8B on real OSINT company profiling: 76.7% vs 0% pass rate, verifiable rewards, receipts…</description>
      <pubDate>Fri, 28 Aug 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/crab1-slm-osint/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>SLM</category>
      <category>QLoRA</category>
      <category>OSINT</category>
      <category>Evaluation</category>
    </item>
    <item>
      <title>Every Improvement I Made to My 1.7B Agent Made It Worse</title>
      <link>https://lobsterswamp.space/blog/small-model-simplicity-wins/</link>
      <description>Four attempts to make a Qwen3-1.7B tool-calling agent smarter. Three made it measurably dumber. The reward bug, the lying eval harness, and the design rule…</description>
      <pubDate>Wed, 05 Aug 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/small-model-simplicity-wins/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>SLM</category>
      <category>Eval</category>
      <category>QLoRA</category>
      <category>Crab-1</category>
    </item>
    <item>
      <title>I Wrote a Self-Healing Script That Could Strand the Machines It Healed</title>
      <link>https://lobsterswamp.space/blog/self-healing-that-strands/</link>
      <description>Renumbering a remote site's LAN from 192.168.1.0/24 to 192.168.2.0/24, with a DHCP safety net that had two bugs. 7 of 10 machines recovered — and the…</description>
      <pubDate>Tue, 04 Aug 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/self-healing-that-strands/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>Networking</category>
      <category>DHCP</category>
      <category>Automation</category>
      <category>Incident</category>
    </item>
    <item>
      <title>Two of My Own Alert Rules Were Wrong. Here Is How I Found Out.</title>
      <link>https://lobsterswamp.space/blog/alert-rules-that-never-fire/</link>
      <description>An alert rule you have never seen fire on a real failure is not a rule, it is a decoration. A temperature rule that screamed at nine idle Macs, a…</description>
      <pubDate>Mon, 03 Aug 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/alert-rules-that-never-fire/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>Alerting</category>
      <category>Prometheus</category>
      <category>Observability</category>
    </item>
    <item>
      <title>A 2012 Mac Mini Beats a Xeon Per Core. That Changed How I Schedule Jobs.</title>
      <link>https://lobsterswamp.space/blog/mac-mini-fleet-benchmark/</link>
      <description>Nine second-hand Mac minis, a used PowerEdge T430 and two containers make a 12-worker, 76-thread cluster. A Nomad sysbatch benchmark showed the dual-Xeon is…</description>
      <pubDate>Sun, 02 Aug 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/mac-mini-fleet-benchmark/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>Nomad</category>
      <category>Benchmarking</category>
      <category>Hardware</category>
      <category>Scheduling</category>
    </item>
    <item>
      <title>Changing a Live Server's Address Changes Its Identity</title>
      <link>https://lobsterswamp.space/blog/nomad-serf-address-identity/</link>
      <description>A 3-server Nomad control plane lost its leader when I flipped the advertise address of every live server at once. In serf/memberlist the address is part of…</description>
      <pubDate>Sat, 01 Aug 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/nomad-serf-address-identity/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>Nomad</category>
      <category>Serf</category>
      <category>Networking</category>
    </item>
    <item>
      <title>Your Codebase Is Lying to You. Here's How to Catch It.</title>
      <link>https://lobsterswamp.space/blog/archdogma-truth/</link>
      <description>90% of linters catch style. ArchDogma catches the moment your codebase becomes a time bomb — with real postmortems, not rule IDs.</description>
      <pubDate>Tue, 19 May 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/archdogma-truth/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>ArchDogma</category>
      <category>Static Analysis</category>
      <category>Python</category>
    </item>
    <item>
      <title>ArchDogma v0.2.0: Scan Your Whole Project, Not Just One File</title>
      <link>https://lobsterswamp.space/blog/archdogma-v020-scan/</link>
      <description>ArchDogma v0.2.0 ships archdogma scan — a whole-project scanner that walks every .py file, runs all Tier 1 detectors, and reports findings with file:line…</description>
      <pubDate>Sat, 09 May 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/archdogma-v020-scan/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>ArchDogma</category>
      <category>Static Analysis</category>
      <category>CLI</category>
    </item>
    <item>
      <title>Kafka Data Flow in a DePIN Pipeline</title>
      <link>https://lobsterswamp.space/blog/kafka-data-flow/</link>
      <description>A visual walkthrough of how QuackNet routes crowdsourced network scan data through Kafka KRaft to ClickHouse, fraud detection, and real-time alerts…</description>
      <pubDate>Fri, 08 May 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/kafka-data-flow/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>Kafka</category>
      <category>DePIN</category>
      <category>Go</category>
    </item>
    <item>
      <title>ArchDogma: Python Architecture Linting That Doesn't Lie</title>
      <link>https://lobsterswamp.space/blog/archdogma-pypi-launch/</link>
      <description>ArchDogma v0.1.0 is now on PyPI. A static analysis tool that catches architectural anti-patterns in Python — not style, not formatting, but structure. 11…</description>
      <pubDate>Thu, 07 May 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/archdogma-pypi-launch/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>ArchDogma</category>
      <category>Python</category>
      <category>Linting</category>
    </item>
    <item>
      <title>I Replaced My Team with 3 AI Agents. Here's the Architecture.</title>
      <link>https://lobsterswamp.space/blog/ai-agent-company/</link>
      <description>How I built a persistent multi-agent AI workforce: Lobster (OpenClaw/Claude), Hermes (CLI agent), and Paperclip as the control plane. Heartbeat protocol…</description>
      <pubDate>Wed, 06 May 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/ai-agent-company/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>Agents</category>
      <category>Orchestration</category>
      <category>Memory</category>
    </item>
    <item>
      <title>I Write Electronic Music in Python: No DAW, No Samples, Just Math</title>
      <link>https://lobsterswamp.space/blog/ai-music-with-numpy/</link>
      <description>How I generate techno and ambient tracks using pure NumPy and SciPy — no DAW, no samples, no plugins. Every kick drum, synth wave, and chord progression…</description>
      <pubDate>Wed, 06 May 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/ai-music-with-numpy/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>Python</category>
      <category>Audio</category>
      <category>DSP</category>
    </item>
    <item>
      <title>DePIN in 2026: Who Actually Buys Crowdsourced Network Data?</title>
      <link>https://lobsterswamp.space/blog/depin-market-who-buys-data/</link>
      <description>Investigative analysis of the DePIN market in 2026: who actually buys crowdsourced network data, pricing models, and what building QuackNet taught me about…</description>
      <pubDate>Tue, 31 Mar 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/depin-market-who-buys-data/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>DePIN</category>
      <category>Market</category>
      <category>Pricing</category>
    </item>
    <item>
      <title>I Benchmarked 7 On-Device LLMs So You Don't Have To</title>
      <link>https://lobsterswamp.space/blog/on-device-llm-benchmark-2026/</link>
      <description>Real-world benchmarking of on-device LLMs on Apple M1: Phi-4-mini, Qwen2.5-3B, Llama-3.2-3B, and more. Discover which model wins for local inference in 2026.</description>
      <pubDate>Tue, 31 Mar 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/on-device-llm-benchmark-2026/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>On-device AI</category>
      <category>Benchmarking</category>
      <category>LLM</category>
    </item>
    <item>
      <title>Prompting Without Pain: What Actually Works in March 2026</title>
      <link>https://lobsterswamp.space/blog/prompting-investigation-2026/</link>
      <description>A practical investigation into prompting techniques for Claude, Grok Imagine, and ChatGPT in 2026. Vector spaces, positive framing, XML tags, and 20 real…</description>
      <pubDate>Tue, 31 Mar 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/prompting-investigation-2026/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>Prompting</category>
      <category>LLM</category>
      <category>Evaluation</category>
    </item>
    <item>
      <title>5 Things Nobody Tells You About Solana Token-2022</title>
      <link>https://lobsterswamp.space/blog/solana-token2022-gotchas/</link>
      <description>Building on-chain token programs with Anchor? Token-2022 is powerful but full of hidden gotchas. Here's what I learned deploying 5 programs for QuackNet.</description>
      <pubDate>Tue, 31 Mar 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/solana-token2022-gotchas/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>Solana</category>
      <category>Token-2022</category>
      <category>Anchor</category>
    </item>
    <item>
      <title>30 Sessions, 1 Developer, 0 Employees: Building a Full-Stack DePIN Platform with Claude</title>
      <link>https://lobsterswamp.space/blog/solo-dev-with-claude/</link>
      <description>A candid builder story: how I built QuackNet, a full-stack DePIN platform with 5 Solana programs, a Flutter app, Go backend, Kafka pipeline, and ClickHouse…</description>
      <pubDate>Tue, 31 Mar 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/solo-dev-with-claude/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>Claude</category>
      <category>Agents</category>
      <category>DePIN</category>
    </item>
    <item>
      <title>How I Built This Portfolio: Zero Dependencies, 11 Architecture Layers</title>
      <link>https://lobsterswamp.space/blog/building-this-portfolio/</link>
      <description>Technical deep dive into building a portfolio with vanilla JS, DDD architecture, EventBus pattern, Three.js particles, WebLLM integration, and a…</description>
      <pubDate>Mon, 30 Mar 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/building-this-portfolio/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>DDD</category>
      <category>Web</category>
      <category>Performance</category>
    </item>
    <item>
      <title>Rust to WASM: Building a Fraud Detection Engine for the Browser</title>
      <link>https://lobsterswamp.space/blog/rust-wasm-browser-fraud/</link>
      <description>A dated correction. I claimed a 6.3 KB Rust/WASM engine at 50,000 scores per second and 10x faster than JavaScript. I withdrew it, then hand-wrote 218 bytes…</description>
      <pubDate>Sat, 28 Mar 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/rust-wasm-browser-fraud/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>Rust</category>
      <category>WASM</category>
      <category>Benchmarking</category>
    </item>
    <item>
      <title>ClickHouse vs TimescaleDB for Geospatial Data</title>
      <link>https://lobsterswamp.space/blog/clickhouse-vs-timescale/</link>
      <description>Why I run both PostgreSQL+ClickHouse in production. Architecture decisions, H3 geospatial indexing, and real query performance numbers.</description>
      <pubDate>Sat, 21 Mar 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/clickhouse-vs-timescale/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>ClickHouse</category>
      <category>TimescaleDB</category>
      <category>Geospatial</category>
    </item>
    <item>
      <title>Running 3.8B Models on 8GB RAM</title>
      <link>https://lobsterswamp.space/blog/phi4-mini-8gb/</link>
      <description>Microsoft Phi-4-mini, MLX framework, 4-bit quantization. How to run capable language models on consumer hardware with real benchmarks.</description>
      <pubDate>Sat, 21 Mar 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/phi4-mini-8gb/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>LLM</category>
      <category>Quantisation</category>
      <category>MLX</category>
    </item>
    <item>
      <title>Kafka KRaft for Solo Projects</title>
      <link>https://lobsterswamp.space/blog/kafka-kraft-solo/</link>
      <description>Why Kafka KRaft mode makes sense for solo developers. Single-process broker+controller, no ZooKeeper, and the gotchas I learned.</description>
      <pubDate>Fri, 20 Mar 2026 00:00:00 +0000</pubDate>
      <guid isPermaLink="true">https://lobsterswamp.space/blog/kafka-kraft-solo/</guid>
      <dc:creator>Yegor Gaidar</dc:creator>
      <category>Kafka</category>
      <category>KRaft</category>
      <category>Docker</category>
    </item>
  </channel>
</rss>
