Thoughts · Notes · Deep Dives
Cursor's swarm-vs-single-agent SQLite experiment is really a context-window management story. A close read of what worked, what didn't, and what it means for memory infrastructure.
A deep dive into the architecture behind Plebz AI's low-latency inference engine — from model quantization to memory-aware scheduling.
How we built cross-application context synchronization at Maximem AI, enabling sub-300ms retrieval in memory-rich pipelines.
My journey contributing to Chromium — creating a cross-platform printing abstraction for Android from scratch.