<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0"><channel><title>Rohit Kushwaha</title><link>https://rohitkushwaha.me/</link><description>Rohit Kushwaha is a Senior AI Engineer and former Tech Lead in Pune who builds software products end to end — interfaces, distributed backends, AI and infrastructure.</description><item><title>OpenDynamicGGUF: An Open, Measured, Reproducible Optimizer for Per-Tensor GGUF Quantization</title><link>https://medium.com/@imrohitkushwaha2001/opendynamicgguf-an-open-measured-reproducible-optimizer-for-per-tensor-gguf-quantization-6ca130764ee4</link><guid>https://medium.com/@imrohitkushwaha2001/opendynamicgguf-an-open-measured-reproducible-optimizer-for-per-tensor-gguf-quantization-6ca130764ee4</guid><pubDate>Sat, 12 Sep 2026 09:00:00 +0000</pubDate><description>Why uniform Q4_K_M is leaving quality on the table — and how an open 15-step pipeline picks the right bits for every tensor, under your hardware budget.</description></item><item><title>Stop Writing Nested Try/Except Hell: The AWS Service That Replaces Your Entire Workflow Engine</title><link>https://medium.com/@imrohitkushwaha2001/stop-writing-nested-try-except-hell-the-aws-service-that-replaces-your-entire-workflow-engine-c992834fdabb</link><guid>https://medium.com/@imrohitkushwaha2001/stop-writing-nested-try-except-hell-the-aws-service-that-replaces-your-entire-workflow-engine-c992834fdabb</guid><pubDate>Sat, 05 Sep 2026 09:00:00 +0000</pubDate><description>If your Lambda functions call other Lambda functions wrapped in retry logic wrapped in more retry logic, you're solving a problem AWS already solved for you.</description></item><item><title>The Data Paradox No One Talks About: Why AI Companies Are Secretly Building Fake Enterprises to Train Real Models</title><link>https://medium.com/@imrohitkushwaha2001/the-data-paradox-no-one-talks-about-why-ai-companies-are-secretly-building-fake-enterprises-to-b1330320f19b</link><guid>https://medium.com/@imrohitkushwaha2001/the-data-paradox-no-one-talks-about-why-ai-companies-are-secretly-building-fake-enterprises-to-b1330320f19b</guid><pubDate>Fri, 04 Sep 2026 09:00:00 +0000</pubDate><description>You need sensitive data to train a system that detects sensitive data — but you're legally forbidden from using it. Here's the workaround nobody's talking about.</description></item><item><title>Airbnb Changed Login Completely — The Server Now Decides How You Authenticate</title><link>https://medium.com/@imrohitkushwaha2001/airbnb-changed-login-completely-the-server-now-decides-how-you-authenticate-6375f903476c</link><guid>https://medium.com/@imrohitkushwaha2001/airbnb-changed-login-completely-the-server-now-decides-how-you-authenticate-6375f903476c</guid><pubDate>Sun, 30 Aug 2026 09:00:00 +0000</pubDate><description>How Airbnb's engineering team scrapped hardcoded login logic, moved the brain to the server, and cut client code by 60% in the process.</description></item><item><title>Spot Instances: The Secret to Cutting Cloud Costs by Up to 90% (And Why Every Modern Team Should Care)</title><link>https://medium.com/@imrohitkushwaha2001/spot-instances-the-secret-to-cutting-cloud-costs-by-up-to-90-and-why-every-modern-team-should-c2a8f1b90e6e</link><guid>https://medium.com/@imrohitkushwaha2001/spot-instances-the-secret-to-cutting-cloud-costs-by-up-to-90-and-why-every-modern-team-should-c2a8f1b90e6e</guid><pubDate>Tue, 25 Aug 2026 09:00:00 +0000</pubDate><description>Cloud computing changed how we build and scale applications — and introduced runaway costs. How Spot capacity cuts the bill by up to 90%, and how to use it safely.</description></item><item><title>I Tried Meta’s “Segment Anything Model”— And Now I Can’t Look at Photos the Same Way</title><link>https://medium.com/@imrohitkushwaha2001/i-tried-metas-segment-anything-model-and-now-i-can-t-look-at-photos-the-same-way-2e2763904a41</link><guid>https://medium.com/@imrohitkushwaha2001/i-tried-metas-segment-anything-model-and-now-i-can-t-look-at-photos-the-same-way-2e2763904a41</guid><pubDate>Sun, 23 Aug 2026 09:00:00 +0000</pubDate><description>The AI tool that turns any image into perfectly cut-out objects with one click — and why designers, doctors and self-driving engineers are obsessed with it.</description></item><item><title>Why Your Database Wasn’t Built for AI Agents (And How to Fix It)</title><link>https://medium.com/@imrohitkushwaha2001/why-your-database-wasnt-built-for-ai-agents-and-how-to-fix-it-46b8a7418583</link><guid>https://medium.com/@imrohitkushwaha2001/why-your-database-wasnt-built-for-ai-agents-and-how-to-fix-it-46b8a7418583</guid><pubDate>Thu, 06 Aug 2026 09:00:00 +0000</pubDate><description>A practical guide to redesigning database architecture for the agentic era.</description></item><item><title>OpenDynamicGGUF: An Open-Source Answer to Unsloth’s “Dynamic” GGUF Quantization</title><link>https://medium.com/@imrohitkushwaha2001/opendynamicgguf-an-open-source-answer-to-unsloths-dynamic-gguf-quantization-b35342933ef0</link><guid>https://medium.com/@imrohitkushwaha2001/opendynamicgguf-an-open-source-answer-to-unsloths-dynamic-gguf-quantization-b35342933ef0</guid><pubDate>Sun, 26 Jul 2026 09:00:00 +0000</pubDate><description>How a new alpha-stage project makes mixed-precision LLM quantization transparent, reproducible and measurable.</description></item><item><title>LLM Quantization</title><link>https://medium.com/@imrohitkushwaha2001/llm-quantization-07d9f7a0e093</link><guid>https://medium.com/@imrohitkushwaha2001/llm-quantization-07d9f7a0e093</guid><pubDate>Wed, 15 Jul 2026 09:00:00 +0000</pubDate><description>The optimisation that put Llama 3, Qwen, DeepSeek and Mistral on consumer GPUs and laptops — what quantization is, and what names like Q4_K_M actually mean.</description></item></channel></rss>
