<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Erik Steiger</title><description>Portfolio and blog posts mostly about technical things</description><link>https://ersteiger.com/</link><item><title>Clerk Auth via the CLI in Rust</title><link>https://ersteiger.com/posts/clerk-cli-auth/</link><guid isPermaLink="true">https://ersteiger.com/posts/clerk-cli-auth/</guid><description>A step-by-step guide to wiring Clerk’s polished web flows into a headless Rust CLI so users can `mycli auth login`, grab a long-lived JWT, and hit protected endpoints without sacrificing security or sanity.</description><pubDate>Thu, 17 Jul 2025 00:00:00 GMT</pubDate></item><item><title>Europe, please get your incentives straight</title><link>https://ersteiger.com/posts/european-incentives/</link><guid isPermaLink="true">https://ersteiger.com/posts/european-incentives/</guid><description>From our phones to our cloud services, Europe depends on American tech. Here&apos;s why our brightest minds keep choosing Silicon Valley over Stuttgart - and what we can do about it.</description><pubDate>Tue, 17 Jun 2025 00:00:00 GMT</pubDate></item><item><title>Deriving Linear Attention and DeltaNet (Part 1)</title><link>https://ersteiger.com/posts/linear-attention-part-1/</link><guid isPermaLink="true">https://ersteiger.com/posts/linear-attention-part-1/</guid><description>Showing that softmax attention and other sub-quadratic models (linear attention, SSMs, etc.) belong to the same class of models that differ in how they store, optimize, and forget memories.</description><pubDate>Sun, 26 Apr 2026 00:00:00 GMT</pubDate></item><item><title>Non-Linear Attention and Test-Time-Training (Part 2)</title><link>https://ersteiger.com/posts/linear-attention-part-2/</link><guid isPermaLink="true">https://ersteiger.com/posts/linear-attention-part-2/</guid><description>From TTT and Titans to the four axes of MIRAS — reframing the hidden state as a tiny model trained online, and placing every modern sub-quadratic architecture (DeltaNet, Mamba2, Gated DeltaNet, TTT, Titans, KDA, ATLAS) on four design choices.</description><pubDate>Fri, 01 May 2026 00:00:00 GMT</pubDate></item><item><title>Chunkwise linear attention and DeltaNet (Part 3)</title><link>https://ersteiger.com/posts/linear-attention-part-3/</link><guid isPermaLink="true">https://ersteiger.com/posts/linear-attention-part-3/</guid><description>Why per-token recurrences are slow on GPUs, how chunkwise DeltaNet uses the WY representation, and how MQAR shows empirically the improvements of delta rule and gating in fixed-state recurrences.</description><pubDate>Sat, 16 May 2026 00:00:00 GMT</pubDate></item><item><title>Building Docker Hub for Typst Templates</title><link>https://ersteiger.com/posts/papermake-registry/</link><guid isPermaLink="true">https://ersteiger.com/posts/papermake-registry/</guid><description>Escaping CrystalReport. Reinventing PDF Stack and Building Docker Hub for Typst Templates on the way.</description><pubDate>Wed, 02 Jul 2025 00:00:00 GMT</pubDate></item><item><title>How to use Plausible in Remix with a Proxy</title><link>https://ersteiger.com/posts/proxying-plausible/</link><guid isPermaLink="true">https://ersteiger.com/posts/proxying-plausible/</guid><pubDate>Tue, 20 May 2025 00:00:00 GMT</pubDate></item><item><title>Generating 1 Million PDFs in 10 Minutes with Serverless Rust</title><link>https://ersteiger.com/posts/rendering-one-million-pdfs/</link><guid isPermaLink="true">https://ersteiger.com/posts/rendering-one-million-pdfs/</guid><description>How to build a modern and scalable PDF rendering service using AWS Lambda.</description><pubDate>Mon, 21 Apr 2025 00:00:00 GMT</pubDate></item><item><title>Softmax attention is a local constant estimator</title><link>https://ersteiger.com/posts/softmax-attention-local-constant-estimator/</link><guid isPermaLink="true">https://ersteiger.com/posts/softmax-attention-local-constant-estimator/</guid><description>Deriving the view of softmax attention as Nadaraya-Watson kernel regression, why local constant estimators have boundary bias, and how Parallax adds a learned local-linear correction on top of softmax.</description><pubDate>Mon, 01 Jun 2026 00:00:00 GMT</pubDate></item><item><title>Cross-modal search for 1 million Zalando products</title><link>https://ersteiger.com/posts/unvail/why-e-commerce-search-sucks/</link><guid isPermaLink="true">https://ersteiger.com/posts/unvail/why-e-commerce-search-sucks/</guid><description>Exploring how AI and cross-modal search can revolutionize e-commerce search, moving beyond traditional text-based methods to understand visual product features and customer intent.</description><pubDate>Sun, 05 Jan 2025 00:00:00 GMT</pubDate></item><item><title>Benchmarking LLM Inference Engines</title><link>https://ersteiger.com/posts/vllm-vs-max/</link><guid isPermaLink="true">https://ersteiger.com/posts/vllm-vs-max/</guid><description>...</description><pubDate>Mon, 07 Jul 2025 00:00:00 GMT</pubDate></item></channel></rss>