Optimizing NVFP4 Grouped GEMM on Blackwell – Worklog
238 µs to 23.8 µs using tcgen05 MMA, TMA async loads, cluster multicast, etc.
19 min
Flushable SSTables in Pebble
~60% performance improvement when ingested SSTs overlap with the memtable
7 min
Enforcing Robust Type Safety with ESLint
Surfacing ~250 type inconsistencies in Linear's type definitions
5 min
Shipping Linear Drafts
Making drafts a first-class entity throughout Linear
10 min
Concurrent Manual Compactions
Segmenting key space and parallelizing execution, 30% perf improvement
7 min