01 / Developer tooling · static analysis
Skillcheck
Make an agent skill easier to review before it becomes part of a workflow.
Abdallah Arslan · Senior Software Engineer
AI writes code.I make it production-grade.
I build reliable interfaces and the tools to review AI-assisted work. Senior software engineer at Deluxe; maker of browser tools for skill analysis, AI learning, and responsive debugging.
- building production software for the web
- 8+ yrs
- of payment records on R360+
- Millions
- engineers use my AI workflows, across five teams
- 20+
- browser products live: four web apps and one extension
- 5
now · Deluxe · R360+ receivables platform
Atlanta, GA · Remote
Selected work
Useful tools. Inspectable engineering.
Three ways I turn an engineering problem into a working product. Explore the decisions, try the tools, and check the evidence.
02 / Interactive learning · AI systems
AI Demystified
Turn AI concepts into small experiments with visible assumptions and inspectable sources.
03 / Browser extension · responsive debugging
PixelParity
Give a responsive bug report a reproducible viewport, zoom, and display context.
Also building
agent-engineering-rules
Install a compact engineering contract while preserving the repository owner's control.
Quemist
Make chemistry arithmetic inspectable through visible steps, units, and reference material.
HeroChí
Give writers immediate text feedback and make token-budget assumptions visible.
Behind the work
An operator’s view of engineering.
I’m a senior software engineer at Deluxe, with experience in fintech, retail, consulting, and business operations. I care about the interface people use and the engineering decisions that keep it dependable.
Background, experience, and skills →PROFESSIONAL WORK
Engineering at Deluxe
A path from demonstration to customer use, faster builds, and consistent frontend behavior across a large microfrontend environment.
React · TypeScript · .NET · AWS · Harness · Docker
Read the decisions and outcomes →Field notes
What building with LLMs actually looks like
Short, evidence-first write-ups of what I've actually observed building with LLMs — each one with the sample size stated and the ten-minute check to falsify it yourself.
Your rules file is not being read
2 min read
I measured a predecessor engineering-rules setup and found that only 13.2% of routed files were loaded. Rewriting the rules was never going to fix that.
- agents
- evaluation
- context engineering
Five of eighteen runs invented evidence. One refused to.
1 min read
In a batch of runs designed to tempt fabrication, five invented a missing fixture to force a red gate green — and the catch that mattered came from an agent cross-checking primary sources, not from a rule.
- agents
- evaluation
- fabrication
AER v5: a compact contract for complete engineering
4 min read
AER v5 puts complete behavior, engineering judgment, and proportionate verification in a 45-line universal kernel, delivered locally for Claude Code and Codex.
- agents
- engineering judgment
- verification
- release engineering