Newsletter
One insight per week.
No filler.
Each issue covers one idea in production ML optimization in enough depth to actually be useful. Benchmarks, tradeoffs, and the things that textbooks leave out.
Targeted at ML engineers and teams running models in production who want to understand optimization at a deeper level — not just use the default settings.
Free. Weekly. No spam. Unsubscribe anytime.
What to expect
Technical depth
Not "here's how to install PyTorch." Assumes you already know what you're doing.
Benchmark-backed
Claims come with numbers. When they don't, they're flagged as opinion or rule of thumb.
Cross-architecture
CNN techniques applied to Transformers. LLM optimization patterns that borrow from classical compression. Bridges, not silos.