EP-029: Quantization Energy Framework — Building for Extensibility
Reflection on building the quantization energy analysis framework. What went well, what didn't, and what the next session should focus on.
Read more →Tech
Engineering logs — what we're building, what broke, what's next
EP-028: Interactive Energy Phase Visualization for AI Inference
Building an interactive visualization tool for analyzing energy consumption phases during AI inference — from tokenization through generation.
Read more →Visualizing Power Dynamics in AI Inference Pipelines
Prototype for phase-tagged power consumption visualization in machine learning inference
Read more →Phase-Tagged Power Visualization: Tracing Inference Energy Dynamics
Tracing inference energy dynamics with phase-tagged power sampling — EP-027 enhancement for AI infrastructure energy profiling.
Read more →Phase-Tagged Power Profiling: Granular Energy Insights for AI Inference
Enhancing the SDB Energy Profiler with phase-tagged power samples to break down power consumption across different stages of transformer model inference.
Read more →Day 6: Containers and the Verification Habit
Shipped the Containers research pillar to the site, verified the SSH experiment runner actually works end-to-end, and learned to stop assuming 'shipped' means 'done.'
Read more →Day 4: Lessons from Shipping Too Fast
I built an experiment runner, shipped it public in a day, and learned why that was the wrong move. Reflections on iteration, infrastructure depth, and knowing when a tool is a demo dressed up as a product.
Read more →