--- title: "Michael Ivertowski (@mivertowski) | Hashnode" description: "Research Engineer and Engineering Director. Read the latest articles by Michael Ivertowski on Hashnode." url: "https://hashnode.com/@mivertowski" --- @mivertowski # Michael Ivertowski @mivertowski · Zurich, Switzerland · Joined September 2025 Research Engineer and Engineering Director ### [Less hardware, more throughput: DotCompute as carbon-reduced tech](https://mivertowski.hashnode.dev/less-hardware-more-throughput-dotcompute-as-carbon-reduced-tech?utm_source=hashnode&utm_medium=feed) DotCompute started as “make GPU compute sane for .NET”.A close second goal: use the hardware we already paid for instead of buying new toys every quarter. This post is about that second part. Why a GPU framework cares about carbon at all Green softw... 7 min read · [mivertowski.hashnode.dev](https://mivertowski.hashnode.dev/less-hardware-more-throughput-dotcompute-as-carbon-reduced-tech?utm_source=hashnode&utm_medium=feed) [Nov 6, 2025](https://hashnode.com/posts/less-hardware-more-throughput-dotcompute-as-carbon-reduced-tech/690d0ae7abc537856739a349) ### [DotCompute RC2 — Cross-Backend GPU Compute for .NET](https://mivertowski.hashnode.dev/dotcompute-rc2-cross-backend-gpu-compute-for-net?utm_source=hashnode&utm_medium=feed) Another framework? Yes.Another abstraction layer that hates your cache and your startup time? No. DotCompute’s first release candidate is ready. It gives you GPU and CPU acceleration from C# with: One kernel definition → multiple backends (CPU SIMD,... 4 min read · [mivertowski.hashnode.dev](https://mivertowski.hashnode.dev/dotcompute-rc2-cross-backend-gpu-compute-for-net?utm_source=hashnode&utm_medium=feed) [Nov 6, 2025](https://hashnode.com/posts/dotcompute-rc2-cross-backend-gpu-compute-for-net/690c78bcd38c0eedf1f0bad2) ### [The beauty of persistent kernels](https://mivertowski.hashnode.dev/the-beauty-of-persistent-kernels?utm_source=hashnode&utm_medium=feed) At some point you get tired of launching the same GPU kernel a few million times a day just to move a few bytes. RingKernels in DotCompute are the answer to that particular form of self-inflicted pain. Persistent kernels flip the model: launch once, ... 6 min read · [mivertowski.hashnode.dev](https://mivertowski.hashnode.dev/the-beauty-of-persistent-kernels?utm_source=hashnode&utm_medium=feed) [Nov 4, 2025](https://hashnode.com/posts/the-beauty-of-persistent-kernels/690a6a604d3e72e6cb8d398f) ### [CUDA Kernel Execution Debugging Journey](https://mivertowski.hashnode.dev/cuda-kernel-execution-debugging-journey?utm_source=hashnode&utm_medium=feed) Short version: we went from 8/70 passing CUDA tests to a stable, auditable path by fixing NVRTC name resolution, argument marshaling, and unified-memory sync in DotCompute. No mysticism—just careful pointers and fewer foot-guns. TL;DR NVRTC will ha... 4 min read · [mivertowski.hashnode.dev](https://mivertowski.hashnode.dev/cuda-kernel-execution-debugging-journey?utm_source=hashnode&utm_medium=feed) [Sep 4, 2025](https://hashnode.com/posts/cuda-kernel-execution-debugging-journey/68b98831d3f55462281917e9) ### [Event-Sourced State Machines in Orleans](https://mivertowski.hashnode.dev/event-sourced-state-machines-in-orleans?utm_source=hashnode&utm_medium=feed) The good thing about agentic coding is, you can easily awake dead repos out of experiments back to life. Event sourced grain state machines is such an example. Short version: take a clean Stateless state machine, host it as a grain, and let events be... 4 min read · [mivertowski.hashnode.dev](https://mivertowski.hashnode.dev/event-sourced-state-machines-in-orleans?utm_source=hashnode&utm_medium=feed) [Sep 3, 2025](https://hashnode.com/posts/event-sourced-state-machines-in-orleans/68b7f4aa5cea4dd74b2af4fe)