Coverage Areas

Model architectures, routing strategies, inference optimization, and the structural decisions behind frontier systems.

• Mixture-of-Experts Routing at Trillion-Parameter Scale
• Speculative Decoding: Latency Engineering for Production LLMs

GPU clusters, interconnect fabrics, datacenter thermals, and the hardware economics of AI at scale.

• Blackwell NVLink Fabric: Topology That Changes Training Economics
• Liquid Cooling at the Rack: Thermal Limits of AI Factories

Community checkpoints, quantization kernels, fine-tuning stacks, and the open infrastructure layer.

• Open-Weight Kernels: The New Infrastructure Layer

Attention mechanisms, sequence modeling primitives, training dynamics, and the mathematical foundations.

• Attention Is Not Enough: Primitives Behind Modern Sequence Models