Hantz Peter
B. Thermoris

Data scientist – LLM/AI engineer. Since 2022, data science and machine learning in production: anomaly detection on billions of bank transactions, NLP and recommendation systems. Since 2025, LLM systems: RAG decision engines, multi-agent pipelines and the agent harnesses that keep them under control, with human review. Before that, six years running servers and networks.

See my work

Agent harness, live

An AI agent is cleaning a product catalogue. Every action it proposes goes through the harness first: a policy check, your approval when needed, and an audit log. You are the reviewer.

Audit log

Simulated run with sample data. The ideas behind it are in my article Learn the Harness Concepts Before You Learn the Framework.

What I bring

Results

Systems, cloud & DevOps

Six years administering servers, networks and databases for an international program, now applied to cloud deployment of data and AI systems.

Linux
Windows Server
Networking & VoIP
Azure
AWS
Docker
Kubernetes
Git & CI
Bash

Data science & AI

Machine learning and large-scale data pipelines in production since 2022, LLM systems since 2025 (RAG, multi-agent pipelines, agent harnesses), across banking, trading, real estate and consulting.

Python
SQL
PySpark
LangGraph
RAG
Agent harness
MCP
scikit-learn
TensorFlow
MLflow
FastAPI
Power BI

Selected work

Side projects

Writing

All articles on DEV

Path

Let's talk

Happy to talk about data science, LLM engineering and AI in production. Email me or send me a message on LinkedIn.