Christoffer Maintz

Consulting across DevOps, backend, CI/CD, developer experience, automation and AI-assisted development, taking independent technical ownership of concrete solutions across software, infrastructure and development workflows.

• CI/CD pipelines from build to test, staging and production with automated quality gates, releases and deployments across environments (incl. Azure Pipelines).
• Containerisation with Docker and Docker Compose; work on Kubernetes-related infrastructure; Azure (Functions, SQL), infrastructure as code, secrets management and reproducible environments.
• Modernising legacy applications: framework and stack upgrades (incl. legacy Shopware), removing deprecated code and consolidating duplicated logic.
• Developer experience: internal CLI tools, project scaffolding and reproducible local environments; linters, static analysis, tests, hooks, dependency and secret scanning moved as early in the flow as possible.
• AI-assisted development: coding agents with custom skills and hooks, guardrails so agent changes are never accepted uncritically, and cost-aware model routing.
• Mapping use cases and integrating AI into existing business processes to automate manual, repetitive work.

Software-consultancy internship in a client context: production-ready backend services, integrations and automation for business-critical e-commerce systems, primarily the client Illux’s Shopware webshop.

• Built Illux Product AI end to end as sole developer: a Shopware 6 plugin generating multilingual descriptions, SEO, categories and tags from product images via generative-AI APIs behind a provider-agnostic abstraction layer.
• Eliminated manual enrichment for most of a 3,000+ product catalogue at 15-20 minutes per product: 1,000+ hours on the existing catalogue alone. Business-critical and still in daily use.
• Asynchronous processing on Symfony Messenger + RabbitMQ with batching, retry with exponential backoff, idempotency, rate limiting and a full audit trail; configurable confidence model with human-in-the-loop approval.
• Artwork visualisation: shoppers preview art in curated rooms or a photo of their own, composited per scene and streamed live over server-sent events.
• Email template preview platform rendering transactional mail against real orders with client emulation, removing the need for manual test orders.
• Contributed cost-aware model routing to an internal agentic coding harness (Python, LangChain).

One deterministic quality standard - format, lint, types, tests, coverage, security - enforced identically while an AI agent edits, before push and in CI. Reusable GitHub Actions workflows, mise templates and a Claude Code plugin, released as versioned open source.

GitHub Actions · mise · Python · POSIX shell · TypeScript · Semgrep · gitleaks · osv-scanner · ESLint · Spotless · PMD · Claude Code

A GitHub Action that labels every new issue with TypeSafe AI's Jev in milliseconds for fractions of a cent - and escalates only the ones it's unsure about to a human or a real LLM.

TypeScript · GitHub Actions · Octokit · Jev · Cloudflare Workers AI

A TypeScript library that vets every tool call an agent proposes through Jev before it runs - allow, block or hold for a human - so an autonomous agent can't rm -rf your box on a bad hunch.

TypeScript · Node.js · Jev · Vercel AI SDK · LangChain

A CLI that streams JSONL/CSV rows through Jev and returns typed classification and score columns plus a confidence field, flagging low-confidence rows for review.

TypeScript · Node.js · Jev · JSONL · CSV

Electives: Advanced Databases, iOS

Ran daily operations across day and evening shifts: staffing, delegation, employee development and process improvement. Led a waste-reduction initiative worth roughly DKK 1M to the bottom line in its first year.

Shared responsibility for daily operations, planning and staff coordination; onboarding and interviews.