Sculpt your AI stack.
Practical tools, experiments, papers, and shared knowledge for people who want more control over AI — from local LLM inference and optimization to cloud services, generative media, and emerging model capabilities.
Projects & resources
Useful things first. These are current DataSculpting.AI projects, with more tools, research, benchmarks, and experiments planned for the same hub.
Local AI Capacity Calculator
Estimate what model sizes and local AI workloads your hardware can realistically support.
Folderizer for Leonardo.AI
Organize generated media into a workflow that is easier to browse, manage, and reuse.
VisLlama Model Serving
Web-based browser administration for GGUF models served through llama.cpp. Configure, launch, route and monitor inference from one focused control surface.
Profile-Guided MoE Compilation
A Data Sculpting paper exploring PGMC — techniques for shaping mixture-of-experts execution.
Inference Forum
New forum for local AI discussions. Post questions, your ideas, share your projects and knowledge.
A home for hands-on AI.
DataSculpting.AI is centered on understanding and extending what modern AI systems can do: tuning the hardware/software boundary, testing models, building useful utilities, documenting techniques, and connecting practitioners who like to experiment.
Build locally. Learn together. Push capability further.
Follow Data Sculpting for project updates, or join the Local AI Enthusiasts group to compare hardware, software stacks, workflows, and what actually works in practice.