Observe
Instrument agent behavior and production traces with tools such as Langfuse, so your team can see what is actually happening.
Independent AI engineer · Berlin
Agents are easy to demo.
Hard to understand.
I help AI teams see what their agents are actually doing, evaluate them with principled thinking and data, and find the highest-value path from 80% towards 99%.
Consulting
See clearly.
Improve what matters.
I create the observability foundation that makes agent behavior legible, then turn real traces into meaningful evaluations and a focused improvement roadmap.
Book a 30-minute callEvaluate
Turn real interactions into datasets, failure categories, graders, and metrics grounded in user and business value.
Improve
Identify the highest-value failures, focus engineering effort, and establish a measurable loop towards dependable performance.
Ten years in AI and data—including building the observability layer for a chatbot at a company serving six million customers.
An owner, not a pair of hands. Technical depth, direct communication, and total commitment to the outcome.
Capacity
Next opening: November 2026
AUG
Full
SEP
Full
OCT
Limited
NOV
Available
Pricing
Day rates
Rates decrease with committed engagement length.
Have an agent that mostly works—but not reliably enough?
Let’s find the signal