A digital workshop where AI meets automation, experiments take shape, and curiosity drives discovery.
Small AI models like GPT-5.6 Luna and GLM 5.3 are reaching Pareto frontier performance at fraction of the cost. Here is how developers and businesses can leverage them practically.
Read Article →Nvidia's $13B acquisition of Hugging Face merges the world's largest GPU maker with the world's largest AI model hub. Here's what it means for developers, competitors, and the future of open AI.
Read Article →OpenAI's first custom inference chip beats Nvidia Blackwell on performance per watt across nearly all workloads. Here's how Jalapeño compares on architecture, speed, and cost.
Read Article →Xiaomi's new Xring O3 processor matches Apple's cores in single-threaded performance and beats them in multi-threaded execution. With 44MB of cache and 21 execution ports, it signals a silicon power shift that nobody saw coming.
Read Article →A developer built an AI dog companion for Skyrim that understands voice commands, executes multi-step plans, and evolves its personality over time — all running on local hardware with sub-500ms latency.
Read Article →Your local LLM is not broken. Quantization, weak system prompts, and basic inference engines silently degrade quality. Here is what to fix.
Read Article →Explore the lab, check out my writing, or get in touch if you want to collaborate.