Engineer first.
Software developer second.
I don't believe in massive consulting slide decks. I believe in Python scripts, PyTorch models, and PLC integration that actually works on the shop floor.
Simon Kirchebner – independent AI consultant, process engineer by trade.
I have worked in the manufacturing industry for more than 10 years and currently spend my days as a process engineer. I know production from the inside: shift work, quality pressure, data that exists but nobody knows where, and machines that have run for 20 years and will run for another 20.
This background is the reason I started TyrolAI. Most AI consulting is delivered by people who have never stood next to a CNC machine at 2 AM during a quality crisis. I have. That is why my focus is on AI that actually works in daily operations – not in slide decks.
My current focus is on knowledge management with RAG systems. I have built TyrolAI Docs, my own enterprise RAG platform, as the foundation of my knowledge-management projects. I also consult on computer vision, predictive maintenance, and process optimization – drawing on my process-engineering background.
Based in Schwaz, Tyrol. I work with manufacturing companies across Austria and the DACH region – on-site when needed, remote when possible.
TyrolAI Docs
Enterprise RAG platform for industrial companies – on-premise, GDPR-compliant, with Microsoft SSO and document-level security. Based on IBM's OpenRAG, hardened for Austrian industry.
Read the blog post →My Mindset
Pragmatism over Hype
No blockchain, no general AI. I focus on deep learning models trained specifically for your exact industrial bottleneck.
Edge Native
Factory data stays in the factory. I deploy everything on local IPCs for zero latency and absolute data security.
Seamless Integration
AI models are useless if they don't talk to your machines. I build robust OPC-UA bridges to your Siemens & Beckhoff PLCs.
My Arsenal
I use the same open-source frameworks powering global tech giants, strictly tailored and compiled for maximum industrial performance.
PyTorch & TensorRT
Hardware-accelerated AI inference
Docker & Kubernetes
Containerized deployments on Edge IPCs
Python, C++ & Rust
Maximum performance where it matters
import torch
import tensorrt as trt
# Load industrial vision model
model = torch.jit.load("tyrolai_defect_detector.pt")
# Optimize for NVIDIA Jetson Edge
engine = trt.Builder(trt.Logger(trt.Logger.WARNING))
network = engine.create_network(1 << int(trt.NetworkDefinitionCreationFlag.EXPLICIT_BATCH))
# Ready for sub 15ms inference
print("TyrolAI Edge Node initializing...")