Replacing Standard Go JSON with Low-Allocation Alternatives to Overcome Performance Bottlenecks in Large-Scale Systems
The Performance Wall: Why encoding/json Stalls High-Throughput Pipelines When data...
Tag archive
The Performance Wall: Why encoding/json Stalls High-Throughput Pipelines When data...
Every vendor claims the seat is dead and the outcome is king. The filings say something duller and more consequential: the seat stayed, a meter went on top of i
Series: How to Run a Workshop People Remember (8/10) Here's a pattern I've watched in hundreds of...
After five years at HackerRank, I'm moving on. I've been putting off writing this post for weeks,...
A defense of running your AI on your own hardware β not as a hobby, but as a position. The case for local-first models, what the cloud costs you, and why the tr
The Human in the Loop Is Not a Safety Feature
Langchain was the obvious choice when I started building AI tools. Everyone used it. The documentation was everywhere. The abstractions felt clean. Then I
Local-First AI Is a Political Statement Now
We live in a world that celebrates knowledge, yet rarely questions how knowledge is acquired. From...
Analysis of Reflection's 501B Beam model. Explore how this new open-weight technology impacts stock,...
I have a theology degree. I taught myself Python using AI tools. I now run a 44-repo open-source AI toolkit. This is not humblebrag β this is a case study in what AI actually changes about who can build software.
When I started building tools that run entirely on your own hardware, with your own models, no API calls to third parties β I thought I was making a technical choice. It turns out I was making a values statement.