Murat Sari

Serial Entrepreneur, Software Architect & AI Engineer

About

Murat Sari is a serial entrepreneur and software architect who has founded several startups and shipped products across the automotive, gaming, and live-stage visualization industries. He studied computer science at TU Wien and has spent over a decade designing large-scale enterprise systems — from real-time data analytics in the oil industry to patented hardware and software for immersive visual experiences.

From 2012 to 2017 he was a core team member of the Ogre3D open-source 3D engine, where he contributed the HLMS physically-based shading system. He has also worked on hardware prototypes and holds patents related to real-time visualization.

On the software side, Murat works across the full stack with TypeScript, Angular, .NET Core, Node.js/NestJS, and increasingly with Python-based ML stacks including PyTorch. He has years of experience as an enterprise Angular trainer, having taught at the universities of applied sciences in Salzburg and Kapfenberg, where he designed and delivered courses on Angular and software architecture. Today he brings that same pragmatic, production-first teaching approach to AI engineering workshops.

Conference Talks

From the AI Jungle to RAG in a Tab

April 12, 2026Gurgaon, IndiaSlides

From Hours to Minute: An AI Case Study with Sheriff

November 20, 2025AI PolandSlides

Agents All the Way Down: The Agentic Stack

September 30, 2026AI DevCraft · OnlineSlides

Developing with Local LLMs: Reality or Illusion?

September 11, 2026Graz, Austria

Blog Posts

Quantized AI 26/07: Summer Edition

Sep 23, 2026The summer of 2026 was quite eventful for AI. As usual, we got new, powerful models. New terms such as Graph Engineering arrived on the scene, and the ongoing discussion about human code review, yes or no, reached another high point. The main event, though, was something else. The Hugging Face incident triggered a new level of AI doomsday discussion and made it clear that this is not just a marketing gag.

Getting started with Local LLMs for Software Development

Aug 17, 2026Learn how to use local LLMs for software development with predictable costs, privacy and independence, and the quality required for real-world projects.

Building a local-first AI standup bot for Slack

Aug 17, 2026An async standup bot for Slack built with CopilotKit Channels, Mastra, and a local Gemma model — local-first, typed tools, SQLite as the source of truth.

Inside MacroQuest: agent-generated UI in Angular, on a local 12B model

Jul 21, 2026

Soverius AI is now maintaining CopilotKit for Angular

Jul 10, 2026Agentic, generative UI for Angular — a maintained, MIT-licensed package, shown with a small Angular + NgRx demo.

Knowledge in a Browser Tab: Fully Local RAG with Gemma 4

Jun 26, 2026Retrieval-Augmented Generation entirely client-side: with Angular, Transformers.js, WebGPU, and Google's new on-device model Gemma 4 E2B

Build a local AI coding agent from scratch

Jun 11, 2026Build a local AI coding agent from scratch — Gemma 4 on llama.cpp, three tools, one loop — then learn why running it unsandboxed is dangerous and how NVIDIA OpenShell contains it.

Behind the curtain of A2UI: how and why it could work

May 12, 2026Benchmarking 14 local LLM configurations on Google's A2UI generative-UI protocol on a DGX Spark. The cliff between what works and what fails is sharper than expected.

Reviewed Posts

Quantized AI 26/02: Glimpse of Fable, Loop Engineering, Apple Intelligence 3rd Gen

Jun 17, 2026Fable was Anthropic's public Mythos-class model for a few days - then a US export-control order took it offline for everyone. We also look at Loop Engineering (an emerging term) and Apple Intelligence third-generation on-device models at WWDC.

What is an Agent Harness?

May 28, 2026Learn what an Agent Harness is, why it matters, and how it improves LLMs through tools, context management, agent runtimes, guardrails, and intent alignment.

AI Fundamentals: How LLMs Work and Their Limitations

Apr 23, 2026A deep dive into how AI models really work, from pre-training and inference to Chain of Thought, agents, and the emerging field of Harness Engineering.

Highlights

Founded several startups spanning automotive, gaming, and live-stage visualization — created patents in real-time rendering
Ogre3D core team member (2012–2017) — contributed the HLMS physically-based shading system
TypeScript, Angular, .NET Core, Node.js/NestJS, Python/PyTorch — full-stack production expertise
Years of experience as enterprise Angular trainer — lecturer at FH Salzburg & FH Kapfenberg, architecture workshops
Lead Software Architect for real-time analytics in the oil industry

Quick Info

Serial Entrepreneur, Software Architect & AI Engineer

4 talks

8 posts

3 reviewed