Structured output, never prose parsing
Models answer in schema-enforced JSON, so the rest of the system gets typed data instead of guesswork.
Seen in: Illux Product AI →
~/what-i-do/ai
AI enablement
AI earns its place when it removes real, repetitive work and you can trust what it does. I build AI features and agent workflows with structure, measurement and humans on the uncertain cases.
01How I work
Confidence decides who reviews
Confident results flow through; uncertain ones go to a person. Automation where it’s safe, judgement where it matters.
Seen in: jev-triage →
Cheapest model that’s good enough
Fast, cheap models handle the bulk; stronger ones are reserved for what they can’t decide.
Seen in: jev-sort →
Measure before trusting
Evaluation harnesses compare models and prompts, audit trails record exactly what ran, and guardrails stop agents before risky actions.
Seen in: jev-guard →
02What I’ve built
Illux Product AI
Generative-AI product enrichment & room visualization for Shopware 6
- PHP 8.1+
- Shopware 6
- Symfony Messenger
- RabbitMQ
Foundry
A polyglot engineering gate for AI-assisted development
- GitHub Actions
- mise
- Python
- POSIX shell
jev-triage
Near-free GitHub issue triage with confidence-gated escalation
- TypeScript
- GitHub Actions
- Octokit
- Jev
jev-guard
A fail-safe guardrail for LLM agents' tool calls
- TypeScript
- Node.js
- Jev
- Vercel AI SDK
jev-sort
jq for judgment - bulk classification at pennies per 10k rows
- TypeScript
- Node.js
- Jev
- JSONL
Jobbuddy
AI job-application platform - from posting to tailored CV
- Java
- Spring Boot
- Angular
- PostgreSQL
Tech Atlas
A bilingual technical dictionary you can explore as a map
- Astro
- TypeScript
- Supabase
- Cytoscape
03What I can do for you
from 9,000 DKK fixed
AI integrations
LLM features that survive production: schema-enforced output, retries and idempotency, confidence scoring, human-in-the-loop review and an audit trail. Any major LLM provider, behind an abstraction you can swap.
from 4,500 DKK fixed
AI-assisted development enablement
Get real leverage from coding agents without the chaos: skills, hooks and a gate the agent can’t quietly weaken, plus workflows from ticket to reviewable PR.
04Writing
One rule set, three placements
Why I stopped treating CI and "AI coding standards" as two separate things - and what Foundry does instead.