Archive
7 posts published so far
Why Forced JSON Output Is Always Valid and Sometimes Wrong
Theaicommit
Why Split Prefill and Decode GPUs Lose Time in the Hand-Off, Not the Network
Theaicommit
Designing a per-user AI cost meter: idempotency, cache pricing and aliases
Theaicommit
Why Speculative Decoding Slows Hybrid Mamba Models at Large Batch Sizes
Theaicommit
EngAI: AI News for Software Engineers, One Swipe at a Time
Theaicommit
How Sorting Your Prompts Speeds Up a Mixture-of-Experts Batch Job
Theaicommit
How to Make Your Agent Check the Records It Changed Before It Says Done
Theaicommit