Stop Asking It to Judge: Use the LLM to Read, Score in Code
Use the LLM only to copy out résumé facts and score them in code. For nine of eleven models the score stopped moving on demographic edits.
Bogdan Szabo
Bogdan is a senior engineer who has spent more than 15 years building software, developer tools and open-source projects. At re:cinq, he works hands-on with agentic coding, AI agents and MCP, building and testing new ways of developing software with AI. His recent work includes research into AI hiring bias and tools for autonomously tackling legacy codebases.
7 Articles • Research • AI • Engineering
The Prompt Won't Save You: Three Objections, Three Tests
Three Reddit objections to our hiring-bias study, run as experiments. A placebo control showed a blank line moves the score as much as a school does.
The Frontier Isn't Special: Claude Fable 5 and Hiring Bias
Anthropic's most capable model lands mid-pack on résumé-screening bias. Frontier capability neither fixes nor worsens silent hiring bias.
The Blind Spot in the Machine: What 25,500 LLM Evaluations Reveal About AI Hiring Bias
An audit of 25,500 LLM resume evaluations across ten models tied nearly half the score differences to demographic bias, not work experience.
I taught an AI my photography style on a $250 GPU
How I trained a custom diffusion model on 100,000 photos with one Intel Arc B580 in nine days, for 21 euros of power, no cloud GPUs.
Simple Tools, Smarter Agents
Why mirroring a REST API into MCP tools suffocates AI agents — and how switching to dynamic, lazy-loaded tools cut context usage by 98%.
The "Ralph Wiggum Method": Why I Built an Infinite Loop on Purpose
A checklist-driven while(true) loop that uses an LLM to chew through documentation, refactors, and tech-debt tickets—reliably and repeatably.
9 min
14 Sept 2026Showing 7 of 7 articles