RGRost Glukhovinrosluk.hashnode.dev·3d ago · 8 min readI Tested ROCm vs Vulkan for AMD Local LLMs — One Backend Silently Tanked by 6xThe same model, the same AMD GPU, two different compute stacks. One hit 143 tokens per second. The other — on a slightly bigger model — quietly collapsed to a sixth of its neighbour's speed, and the s00
RGRost Glukhovinrosluk.hashnode.dev·Aug 7 · 10 min readYour LLM Guardrails Don't Protect Your Agents — Here's What DoesYou've set up input validation. You're filtering outputs. Your model is sandboxed behind guardrails. Now your agents are calling tools, delegating to other agents, and acting on user data — and none o00
RGRost Glukhovinrosluk.hashnode.dev·Jul 2 · 10 min readGoogle's A2A Protocol in 2026: What Actually WorksEveryone told me the Agent-to-Agent protocol was premature. Then Google announced it had crossed 150 supporting organizations and landed in major cloud platforms. The numbers looked impressive. The re00
RGRost Glukhovinrosluk.hashnode.dev·Jan 24 · 3 min readGPU and RAM Prices Surge in Australia: RTX 5090 Up 15%, RAM Up 38% - January 2026Today we are looking at the top-level consumer GPUs, and RAM modules. Specifically I’m looking at RTX-5080 and RTX-5090 prices, and 32GB (2x16GB) DDR5 6000. I’ve taken screenshots of cheapest prices for in-stock graphic cards on the three prominent ...00
RGRost Glukhovinrosluk.hashnode.dev·Jan 10 · 1 min readIT Events in Melbourne in 2026Melbourne continues to be a vibrant hub for technology professionals in 2026, offering a diverse range of conferences, workshops, and community meetups. For people interested in software development, cybersecurity, cloud computing, or AI, Melbourne’...00
