<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Sridhar Vanka — Writing</title><link>https://sridharvanka.me/writing</link><description>Essays, experiments and field guides by Sridhar Vanka.</description><language>en</language><atom:link href="https://sridharvanka.me/feed.xml" rel="self" type="application/rss+xml"/><item><title>Checking when a coding agent is done</title><link>https://sridharvanka.me/coding-agent-eval</link><guid isPermaLink="true">https://sridharvanka.me/coding-agent-eval</guid><description>What twenty local-model runs revealed about completion checks, the evaluator, and the cost of trying again.</description><pubDate>Sun, 11 Oct 2026 00:00:00 GMT</pubDate></item>
<item><title>Can AI help me find information in my saved screenshots?</title><link>https://sridharvanka.me/screenshot-recall</link><guid isPermaLink="true">https://sridharvanka.me/screenshot-recall</guid><description>Why I’m testing local screenshot search, the failures that changed the design, and what remains to measure on the phone.</description><pubDate>Sun, 11 Oct 2026 00:00:00 GMT</pubDate></item>
<item><title>Looking beyond LLMs</title><link>https://sridharvanka.me/outside-llm</link><guid isPermaLink="true">https://sridharvanka.me/outside-llm</guid><description>Local case studies comparing a language model with classical ML and specialized models across different kinds of data.</description></item>
<item><title>Does code still need to be maintainable if AI maintains it?</title><link>https://maintainability.sridharvanka.me/</link><guid isPermaLink="true">https://maintainability.sridharvanka.me/</guid><description>A blinded replication examines repeated maintenance, task success, retries, and cost.</description></item>
<item><title>The Mixed Workforce Company</title><link>https://sridharvanka.me/research/mixed-workforce-company/</link><guid isPermaLink="true">https://sridharvanka.me/research/mixed-workforce-company/</guid><description>An interactive model for sharing work, authority, and accountability between people and AI.</description></item>
<item><title>The Human Exclusion Threshold</title><link>https://human-exclusion.sridharvanka.me/</link><guid isPermaLink="true">https://human-exclusion.sridharvanka.me/</guid><description>When speed and complexity make intervention impractical, where should human authority sit?</description></item>
<item><title>The Right to Be Forgotten Needs Rewriting</title><link>https://sridharvanka.me/research/unlearning/</link><guid isPermaLink="true">https://sridharvanka.me/research/unlearning/</guid><description>What deletion means when information is distributed through a trained model.</description></item>
<item><title>What an AI practice is actually made of</title><link>https://sridharvanka.me/ai-practice/</link><guid isPermaLink="true">https://sridharvanka.me/ai-practice/</guid><description>Thirteen areas of an AI practice, with proposed field tests and the evidence each would need.</description></item>
<item><title>How much can a computer do with borrowed energy?</title><link>https://sridharvanka.me/research/ambient-iot</link><guid isPermaLink="true">https://sridharvanka.me/research/ambient-iot</guid><description>Battery-free shipment tags, the limits of harvested energy, and models you can try.</description></item>
<item><title>The Ten-Minute Household</title><link>https://sridharvanka.me/india/quick-commerce/</link><guid isPermaLink="true">https://sridharvanka.me/india/quick-commerce/</guid><description>What moves from the household to the network when delivery takes ten minutes?</description></item>
<item><title>UPI: The Price of Free</title><link>https://sridharvanka.me/india/upi/</link><guid isPermaLink="true">https://sridharvanka.me/india/upi/</guid><description>Follow a payment through the institutions, incentives, costs, and risks behind the interface.</description></item></channel></rss>
