Our Mission
AI is a new attack surface. At the same time, cloud AI is a data sovereignty problem. GuCoIA combines security engineering with local AI infrastructure: sensitive data stays on-premise, requests are intelligently routed by sensitivity level, and every request is automatically checked against security policies.
Founder
Mike Thieke
Mike Thieke has extensive experience in designing and operating secure architectures for international projects. As founder of GuCoIA, he combines deep technical expertise with practical security governance experience.
Professional Background
- Cyber Security Manager at Leopold Kostal GmbH & Co. KG (01/2024 - present)
- Lead domain security for international projects across three continents
- Coordinate teams of 50 engineers on ISO 21434 compliance
- Drive threat modeling, risk assessments, and security governance
- Supervise penetration tests, fuzzing campaigns, and static code analysis
- Developer at Materna Information & Communications SE (06/2017 - 11/2022)
- Innovation Center Blockchain Team (Hyperledger Fabric, Ethereum security)
- Implemented micro-service security patterns and Kubernetes service-mesh (Istio)
- Managed Helm-based deployments and Infrastructure-as-Code with Terraform
- Applied Machine Learning for object-recognition prototypes
Education
- M. Sc. Cyber-Security, Ruhr-Universität Bochum (2020-2022)
- Thesis: Comprehensive Analysis of User Namespaces via Fuzzing
- B. Sc. Cyber-Security, Ruhr-Universität Bochum (2016-2020)
- Thesis: Sender Identification for CAN-FD Networks
Certifications
- ASPICE Sec-Pam (Automotive Security Process Assessment)
- Scrum Master (PSM1)
Expertise
- Security: ISO 21434, Threat Modeling, Penetration Testing, Fuzzing
- Cloud: Kubernetes, AWS, IBM Cloud, Terraform, Helm
- Programming: C, C++, Java, Spring Boot, Node.js, React
- Methods: Scrum, ASPICE, Service-mesh (Istio)
Languages
- German (Native)
- English (Fluent, C1/C2)
- French (Good command, B1/B2)
- Spanish (Basic, A2)
Contact
- Sankt Augustin, Germany
- kontakt@gucoia.de
- +49 152 31 38 67 94
The Three Pillars
AI Security Consulting
OWASP LLM Top 10 assessments, threat modeling, and KI Security Firewall design:
- Prompt Injection Hardening: Direct and indirect attack mitigation
- KI Security Firewall Design: Sensitivity-based routing, enforcement layer, audit trail
- Threat Modeling: STRIDE analysis for LLM attack surfaces
- Incident Response: Readiness planning for AI security events
Local AI Infrastructure
From assessment to production-ready deployment:
- Infrastructure Assessment: Hardware sizing, GPU/NPU requirements, TCO analysis
- Edge AI Deployment: llama.cpp, vLLM, LoRA — air-gapped or hybrid
- Performance Optimization: KV-cache sharding, async precomputation, late chunking
- Security Monitoring Retainer: Ongoing hardening and patch tracking
KI Security Firewall (Product)
Sensitivity-based model routing and security enforcement:
- Smart Routing: PII/confidential → local models, general → EU cloud
- Enforcement Layer: Prompt injection detection, PII filter, policy enforcement
- Audit Trail: Full request logging for compliance evidence
- Model Lineage Tracking: Which model saw which data, when
Our Values
Security-First
AI security as core infrastructure, not compliance checkbox
Sovereignty
Sensitive data stays on-premise, by architecture
Transparency
No black boxes: audit trails, model lineage, open-weight models
Mid-market Focus
Enterprise-grade security at SME-accessible prices
Contact Us
Ready to secure your AI journey? Get in touch.