Projects – Alexander Becker

01WORKSHOP · ARTIFICIAL INTELLIGENCE

AI in everyday school life – a workshop for teachers

How do language models work, how can they be used sensibly in the classroom – and how should schools respond to the fact that students have long been using AI? A practical, in-person workshop for the teaching staff of a grammar school.

Goals

1

Understand

Basic principles of AI and language models – no technical background needed.

2

Apply

Use free AI tools for lesson preparation and creating materials.

3

Assess

Critically reflect on the opportunities and risks of student use.

Structure

01
AI fundamentalsTerms, history, neural networks, strengths, weaknesses and data protection
02
AI in the teaching professionFields of use, subject-specific examples, tool overview and live demo
03
Hands-on: try it yourselfCreate and review your own teaching material in small groups
04
Students and AIUsage patterns, limits of AI detectors, legal framework, communicating with parents
05
Hands-on: designing AI-resistant assignmentsRework existing assignments – or deliberately integrate AI
06
First school-wide guidelinesKey points for a school-wide approach to AI

02PODCAST · TEST MANAGEMENT

Test Data Management meets Data Science

On the Qytera podcast I talk about how data science methods improve test data management – from choosing between synthetic and production data to the targeted reduction of large data sets.

What it's about

Synthetic data and production data each have their own strengths depending on the type and phase of testing. What matters is knowing the test-relevant data constellations – instead of blindly using complete production copies.

To do this, I classify test data into three categories of data rules and show how data science helps derive them from real data:

1

Technical rules

Formats, keys and structures the data must satisfy.

2

Data quality

Completeness, plausibility and consistency of the data.

3

“Reality rules”

How data is actually used – the basis for risk-based testing.

Methods discussed

01
Usage analysis of real dataRegression analyses to reveal relationships and frequencies
02
Test data reduction with clusteringIdentify representative records instead of testing millions of cases
03
Machine learning in practiceEffort, algorithms used and time saved with large data volumes

03WORKSHOP & STRATEGY CONSULTING · ARTIFICIAL INTELLIGENCE

From introducing AI to your own company RAG

A workshop with compact strategy consulting for mid-sized companies: what can AI do for knowledge work in your own organisation – and how can you build an AI solution on internal data without giving up control over it?

What AI does for knowledge work

AI is a general-purpose technology – like electricity or the internet, its value only unfolds in application. For companies, three capabilities count:

1

Transform

Turn knowledge from meetings, documents or images into new formats – implicit knowledge becomes explicit.

2

Connect

Link internal and external sources by semantic similarity rather than exact matches.

3

Automate

Automate processes with workflow tools far more easily than before.

01
Example workflowFrom a recorded meeting via transcript and summary to tasks, minutes and a knowledge base
02
Data as a fundamental questionWhich data do I want to share with whom – between data protection, data autonomy and added value
03
Architecture building blocksInterface, models, workflow automation, embedding, vector database and integration of existing services
04
Open-source stackE.g. Open WebUI, Ollama with local models, n8n and PostgreSQL as a vector database
05
RAG pipelineHow retrieval-augmented generation makes company knowledge usable for language models

Sharing knowledge