AI Compass
Compass

MLOps

What lies between a working notebook and dependable operation: versioning, monitoring, drift, rollback and evidence.

·2 min read·By Fachredaktion Technik
DETAIL
3 sections

The idea

Training a model is an experiment. Running one is a commitment: that it will do the same thing tomorrow, that someone is responsible, and that you can show what it did and when.

The four questions

QuestionAnswer in operation
Which model is running right now?Model registry with version and checksum
What was it built from?Data, code and configuration state
How good is it today?Continuous measurement against a fixed set
What do we do when it worsens?Trigger, rollback, ownership

Detecting drift

TypeWhat changesHow measurable
Data driftThe inputsDistribution comparison per feature
Concept driftThe relationshipOnly with delayed labels
Label driftThe class balanceProportions over time
Prediction driftThe outputsDistribution of model outputs

The fourth row is the only one available immediately and therefore the practical entry point: a shift in the output distribution is an early warning long before labels arrive.

import numpy as np

def psi(expected, observed, bins=10):
    """Population stability index. Above 0.2 counts as a clear shift."""
    edges = np.percentile(expected, np.linspace(0, 100, bins + 1))
    edges[0], edges[-1] = -np.inf, np.inf
    e = np.histogram(expected, bins=edges)[0] / len(expected)
    o = np.histogram(observed, bins=edges)[0] / len(observed)
    e, o = np.clip(e, 1e-6, None), np.clip(o, 1e-6, None)
    return float(((o - e) * np.log(o / e)).sum())
  • PSI below 0.1: unremarkable. 0.1 to 0.2: watch. Above 0.2: investigate.
  • Monitor not only features but the share of missing values per column.
  • Run a shadow deployment before a new model goes live.

What makes a run reproducible

  • Data: a snapshot with a checksum, not just a path. A directory changes.
  • Code: a commit identifier, not a branch name.
  • Environment: the container image digest, not a package list.
  • Configuration: complete, including seeds.
  • Result: model checksum and metrics on a fixed evaluation set.

Those five items are simultaneously what Annex IV of the AI Act requires as technical documentation for high-risk systems. Keeping them anyway discharges most of that obligation. See Preparing for audit.

The rollback

A rollback is only an option once three conditions hold:

  1. 01

    The previous version is available

    Not only the weights but the complete serving environment.

  2. 02

    The interface is compatible

    Rollbacks routinely fail because the output format changed and downstream systems expect the new one.

  3. 03

    It has been rehearsed

    With a measured duration. An unrehearsed rollback takes about three times the estimate when it matters.

The metrics that actually get read

MetricWhy
Share of refused or escalated casesEarly warning, available immediately
Time to first token, 95th percentileUser experience, not the mean
Share of answers with a citationFor RAG the decisive figure
Human correction rateThe most honest quality figure there is
Cost per completed taskThe only number the business compares

The fourth row is the most valuable and the least often collected: if a human is reviewing the result anyway, their correction is a free label. Systematically collecting those corrections yields, within six months, a dataset no vendor can supply.

Was this page helpful?
MLOps