Viewing profile — tweedler290

Token cost and wall clock time are practical pain points for SWEs, but the bigger intellectual pin point is that the dang bot has to reconstruct the code structure actually READING…

I made a code intelligence system from scratch! language agnostic and model agnostic. runs locally. Benzi is a compiler+harness+runtime tracer/debugger built from ground up to writ…

This is really cool! Had a question on the MoE: in kimmi-K2/model.py, the router does torch.topk(..., k=self.num_experts) while every other MoE uses k=self.top_k. The ctor's top_k=…

thank you!! and it scales. the same operation runs no matter the size of repo. once you have paid the cold start price, it updates incrementally (AI or user writing code).

harness is only available as a VS extention for now (no headless. MCP release soon as well), you can get it here: https://marketplace.visualstudio.com/items?itemName=variantt... he…

Roughly speaking, the way current AI coding agents/harnesses work is by either: a) Pulling in appropriate text snippets of code across multiple files and handing them to the agent,…