Daniel Kreuzhofer

AWSGenAILLM AgentsFounder

Cloud Architect & AI Solutions Expert

Learn More

About

Cloud Architect & AI Solutions Expert

I'm Daniel Kreuzhofer — 20+ years in IT, from building trading platforms at UniCredit to leading cloud architecture teams at AWS. Today I help companies in retail, finance, and insurance turn AI from buzzword into business results. Whether it's migrating to AWS, modernizing legacy apps, or implementing GenAI that actually works, I focus on practical outcomes over hype. I bring the technical depth of a solutions architect with the practical mindset of a founder — I've built companies, shipped products, and know what it takes to go from idea to execution. Based in Bavaria, Germany.

I help you move from ideas to real results with AI and cloud — through clear guidance, practical steps, and hands-on coaching.

Experience

Projects

Skills

Proficiency:
Expert
Proficient
Familiar

Cloud Architecture

Designing and implementing scalable cloud solutions on AWS

AWS Solutions Architecture

Expert·8 years

Certified Solutions Architect with deep expertise across AWS services

Cloud Migration

Expert·8 years

Led enterprise migrations for media, healthcare, and entertainment companies

Application Modernization

Expert·6 years

Transformed legacy applications to cloud-native architectures

Microsoft Azure

Proficient·6 years

Certified Azure Solutions Architect, ISV partner onboarding

Infrastructure as Code

Proficient·5 years

CloudFormation, Terraform, and automated deployments

AI & GenAI

Implementing AI-driven solutions for business transformation

Amazon Bedrock

Expert·3 years

Primary AI platform for customer engagements — Claude models, RAG pipelines, agent architectures, and intelligent document processing

GenAI Implementation

Expert·3 years

AWS Certified AI Practitioner. Use-case-first methodology: identify business outcomes, validate feasibility, one-week prototype sprints, then PoC decision

RAG & Hybrid RAG

Expert·2 years

Most common customer pattern — knowledge augmentation for employees using search and reranking for better customer conversations and support

Intelligent Document Processing

Expert·2 years

End-to-end IDP architectures combining Textract and Claude — invoice matching, delivery note mapping, fraud detection. Discovery-first approach for document type, extraction strategy, and data flow design

AI Agents

Expert·2 years

Building agents with Strands Agents SDK (code path) and Amazon Q (no-code path). Deployed agents for non-developer productivity — account managers autonomously creating agents for briefings, reports, and customer communication

Amazon Q

Expert·1 year

No-code AI path for customers — RAG, agents, and MCP integration without deep technical capacity required

Claude Code

Expert·1 year

Primary AI coding tool — built entire Chat3D and portfolio projects using Claude Code CLI with spec-driven workflows, hooks, and MCP server integrations

Kiro

Expert·1 year

IDE and CLI for developer productivity and general employee productivity. MCP server integration for Slack, email, Salesforce, Outlook, OneDrive. Building shared collaboration spaces with Obsidian for agent-readable/writable workflows

Windsurf

Proficient·1 year

First AI coding tool adopted — used for initial project builds before transitioning to Kiro and Claude Code for spec-driven development

ChatGPT / OpenAI Codex

Proficient·1 year

Regular use for code generation, problem-solving, and technical writing. Hands-on experience with GPT-4, GPT-5.2, and Codex models

LLM Fine-tuning

Proficient·1 year

Hands-on fine-tuning of Qwen3-8B, Qwen3-32B (LoRA), and Qwen3-235B-A22B models on GPU clusters. Achieved 2% to 88% accuracy improvement on SQL generation tasks. Experience with FSDP distributed training, LoRA/PEFT, and HuggingFace Transformers

vLLM

Proficient·1 year

High-performance LLM inference serving with OpenAI-compatible API. Multi-node inference with tensor parallelism (TP=8) and pipeline parallelism (PP=2) across 16 H100 GPUs

Ray

Proficient·1 year

Distributed computing framework for multi-node LLM inference coordination. Ray cluster setup across Slurm-managed GPU nodes with InfiniBand + GPUDirect RDMA

Slurm

Proficient·1 year

HPC job scheduler for GPU cluster management. Slurm-on-Kubernetes via Soperator, multi-node training and inference job orchestration on H100 clusters

Machine Learning

Proficient·4 years

ML workloads on SageMaker, model fine-tuning for specific use cases when labelled data is available

Software Development

Full-stack development with modern technologies

TypeScript

Expert·5 years

Full-stack development, React applications, Node.js backends

Python

Proficient·4 years

Automation, AI/ML tooling, scripting

C# / .NET

Expert·15 years

Enterprise applications, ASP.NET, .NET modernization

React

Proficient·4 years

Modern web applications and SPAs

Node.js

Proficient·5 years

Backend services, APIs, serverless functions

Leadership & Strategy

Guiding teams and driving technical strategy

Team Leadership

Proficient·2 years

Led Solutions Architecture teams at AWS

Technical Project Management

Expert·10 years

Enterprise projects in banking, media, and healthcare

Customer Engagement

Expert·15 years

Enterprise customer relationships and technical consulting

AI Strategy & Adoption Consulting

Expert·2 years

Helping enterprises cut through AI hype — filtering 20-30 use cases to the ones with real business impact, consolidating fragmented initiatives, defining measurable outcomes, and managing expectations between management mandates and operational reality

Business Model Transformation

Proficient·4 years

ISV partner consulting and cloud adoption strategies

For Recruiters & Hiring Managers

I believe in radical transparency

Skip the guesswork. Get honest insights into my expertise and an AI-powered assessment of how I match your specific requirements.

Let's Build Something Together

Whether you're looking to modernize your cloud infrastructure, implement AI solutions, or discuss opportunities—I'd love to hear from you. No pressure, just conversation.

No pressure—reach out whenever feels right.