// about
Morris Zimmermann
I'm Morris Zimmermann — 20, based in Hamburg, working as a Junior AI Engineer at Circus Group.
I started in AI at 17 as a working student at FULLY AI. Within a year I was building cloud-based QA infrastructure on GCP and AWS, running red-team exercises on production LLM deployments, and developing the Supervisor AI Framework — a security layer against prompt injection, jailbreaks, and system-message exfiltration. When FULLY AI was acquired by Circus Group, I moved with the team.
My work is technical first. I don't sell slides — I build things. That means writing code, designing systems, and getting into the details of how models actually behave (and misbehave). I'm equally comfortable in a Python script, a Svelte component, a Kubernetes cluster, or with a soldering iron in hand.
Outside work I run Atlas, a modular smart home platform built on MQTT and K3s. More infrastructure than product, which is probably what I enjoy most about it.
Experience
Jun 2025 – present
Junior AI Engineer
Circus Group
Building AI systems for robotics and food-service applications after FULLY AI's acquisition by Circus Group. Technical lead for Circus XR Experience — an AR platform for external pitches. Supporting AI development at CIRCUS SE.
Jul 2024 – Jun 2025
AI Specialist
FULLY AI (acquired by Circus Group)
Developed the Supervisor AI Framework to protect against prompt injection, jailbreaks, and system-message exfiltration. Built a modular prompt engine for non-technical users. Led risk and compliance analyses for customer LLM deployments. Conducted AI penetration tests on internal deployments and prompts.
May 2023 – Jun 2024
Working Student — AI Security & Risk Mitigation
FULLY AI
Built a cloud-based QA center (Python, GCP, AWS) for continuous monitoring of LLM quality, security, and performance. Visualised one/multi-shot test results in Grafana dashboards for real-time quality assurance.
Skills
AI & LLMs
- — Prompt Engineering
- — AI Security / Red-teaming
- — Custom Solutions
- — Infrastructure
Programming
- — Python
- — Flask
- — TensorFlow
- — TypeScript
- — Svelte / SvelteKit
- — C#
- — SQL
Hardware & Embedded
- — Raspberry Pi (Pico, W2, 4)
- — ATmega
- — STM32
- — PCB Design
- — µPython
- — Custom Linux Images
- — MQTT
XR
- — Unity 3D
- — Meta Horizons / Quest
- — CAD Model Optimisation
- — VR & XR Solutions
DevOps & Infra
- — K3s / Kubernetes
- — Docker
- — GCP
- — AWS
- — CircleCI
- — GitHub Actions
AI Dev Tooling
- — Cursor
- — GitHub Copilot
- — OpenAI Codex
How I work
I work best on problems with real constraints. Vague briefs and open-ended sandbox projects don't interest me as much as situations where something actually has to work — reliably, securely, on time. I'm direct, I prefer async communication, and I'd rather deliver something clean and small than something sprawling and half-finished.
Currently exploring
- Agentic AI systems and multi-step tool use
- AR/spatial computing (Unity, Meta Quest, Meta Horizons)
- Low-power embedded Linux (custom images, read-only rootfs)
- K3s and distributed systems at home-lab scale
Languages
German Native
English Fluent (C2)