PinnedInTowards AIbyMustafa Genc·Apr 15Running a 35B Model Locally with TurboQuant — What’s Actually Possible Right NowBefore diving in, one important distinction: TurboQuant does not quantize model weights. It compresses the KV cache at inference time. This…A response icon7A response icon7
PinnedInAI AdvancesbyMustafa Genc·Apr 21Kimi K2.6 Shipped. Palantir Published. The West Is Walking Backwards.A trillion open parameters shipped out of Beijing the same week a $350 billion American defense-tech company posted a 22-point manifesto…A response icon30A response icon30
PinnedMustafa Genc·Feb 15Ollama vs vLLM: A Comprehensive Guide to Local LLM ServingThe Complete Comparison You Need Before Choosing Your LLM Inference EngineA response icon2A response icon2
Mustafa Genc·Nov 6Voters Can Block a Data Center. They Can’t Take Back an AI Model.In the same week, towns across America voted against the buildings AI runs in, and OpenAI put a price of $1,279 on a working attack…
InAI AdvancesbyMustafa Genc·Sep 9Cost-Engineered Agentic Coding: Building a Production-Grade Local LLM Proxy, Multi-Agent…A comprehensive engineering deep-dive into eliminating SaaS vendor lock-in, cutting LLM context costs by up to 80%, and enforcing…
InAI AdvancesbyMustafa Genc·Jul 31Nine Days That Split the AI Industry: A Chinese Model, a Self-Inflicted Hack, and the Letter…A 2.8-trillion-parameter model shipped out of Beijing the same week OpenAI’s own red team accidentally proved how far its models could…
InTowards AIbyMustafa Genc·Jul 27The “Thinking” Penalty: Why Pushing AI to Reason Is Breaking Simple TasksFrom 70x token taxes to degraded accuracy, new benchmarks reveal the hidden costs of Chain-of-Thought — and how to fix them with an…
Mustafa Genc·Jun 23The Death of Token Economics: Why “Cost Per Outcome” is the New Return on Investment (ROI)
InGoPenAIbyMustafa Genc·Jun 14Walls, Not Trust: Building AI Systems That Work Despite Being UnreliableAI systems don't become trustworthy through smarter models. Trust comes from architecture, boundaries, validation, and control.
InTowards AIbyMustafa Genc·Jun 5The Model Was the Easy Part: A Practitioner’s Guide to AI LicensesA practical guide to the legal layer of AI — the one most engineers skip until it costs them.
