Experience

What I work on in production

Mostly one thing: AI voice agents that talk to real customers, and everything that has to work around them.

Now · since 03/2026

Working student, AI applications

I work hands-on on production voice agents built on Leaping, handling real customer calls. I look at what actually happens on calls, then change how the agents and the workflows around them behave.

  • Routing, prompts and conversation behaviour, function/tool nodes and field logic in Leaping.
  • Customer verification, API/CRM-backed actions, tickets and transfers, checked against what the functions actually did.
  • Outbound flows, including a scheduling path that books through a tool and only confirms on success.
  • Production call audits, failure analysis, QA categories and regression cases before changes ship.

Related case studies

Public, sanitized versions: no employer details, prompts, recordings or customer data.

What surrounds one call

Before

Outbound flows · scheduler flows

During

Prompt behavior · function/tool nodes · routing · verification · structured JSON

After

API/CRM actions · post-call workflows · ticket creation

Always

Evaluation · edge-case testing · failure analysis · human review

The main thing it taught me: the hard part of an AI agent isn't what it says. It's everything around it. Which tool fires, what gets written to the CRM, what happens after the call ends, and how you notice when one of those quietly fails.

Before that

  1. 05/2025 – 09/2025

    Assistant

    Medical practice, Berlin

    Scheduled and coordinated patient appointments by phone, and was the first point of contact at reception.

Education

  1. since 10/2025

    B.Sc. Human-Centered Informatics & Psychology

    University of Duisburg-Essen

Have something messy that should be a system?

Start a conversation