Yoni

@yorom24

Software Engineer

Joined May 2025

186 Following

7 Followers

Replies Reposts Media

Pinned
Yoni
@yorom24
Jul 29, 2025
Hi everyone! I'm a Backend Engineer with solid experience in building robust REST APIs, working with Golang,Node js, DJango Postgres, MongoDB and tools like, Mongoose, Gin, Microservice architecture kafka, docker and kubernetes. Feel free to DM. #hireme
Yoni
@yorom24
Sep 24
I built JevPolicy an open-source TypeScript runtime that turns probabilistic AI judgments into deterministic, auditable decisions using policy-as-code. Includes replay, shadow evaluation, semantic diffs, fallbacks, and OpenTelemetry.
GitHub - Sanoy24/jevpolicy: JevPolicy is an open-source TypeScript decision runtime that turns...
From github.com
Yoni
@yorom24
Aug 26
Reviewed the same repo twice once as me, once as a fresh agent with no memory of my findings. It caught a bug I'd read right past. Built this so AI code review can't fake numbers or invent findings. @CodeTrendr
github.com
GitHub - Sanoy24/backend-performance-review: An open-source Agent Skill teaching AI coding agents...
An open-source Agent Skill teaching AI coding agents evidence-based backend performance review — technology-agnostic, workload-driven, invents nothing. - Sanoy24/backend-performance-review
Yoni
@yorom24
May 8
I kept treating preference leakage and self-preference bias as one problem under "cross-family rotation." They are not the same bias. They live at different pipeline layers and require different fixes.
1
Yoni
@yorom24
May 7
You trained a SimPO judge on 600 preference pairs. Reward accuracy hit 1.0 at step 100. Eval loss came down clean. On held-out, your trained model loses to a regex script (30% vs 34%) and to its own untrained backbone by 28pp. What broke? Not the trainer. The data.
1