I'm Jonas Öchsner,

I'm a third-semester Applied Computer Science student at Universität Bamberg, currently building an AI-powered data-protection classifier at the Leibniz Institute for Educational Trajectories — a project I run end to end, from requirements through to maintenance.

Most of my work sits where machine learning meets systems that have to be defensible: models that produce an answer, and an architecture that can explain where the answer came from. I spent this year reproducing a neuro-symbolic reasoning pipeline from the literature, and the rest of it making LLM inference run entirely on-premise so sensitive research data never leaves the building.

From September 2026 I'll be in Dublin on an ERASMUS exchange at TU Dublin.

Get In Touch Download CV

Experience

Software Development for AI-Powered Data Protection Classification

LIfBi — Leibniz-Institut für Bildungsverläufe · Bamberg, Germany

  • Run my own agile-managed software project end to end, from requirements engineering through to maintenance.
  • Ensure the solution complies with European data protection law.
  • Handle customer communication and set up continuous integration for the project.

Oct 2025 — Present

Tutor — "Introduction to AI"

Cognitive Systems Chair, Otto-Friedrich-Universität Bamberg · Bamberg, Germany

  • Teach and support students in the "Introduction to AI" course.

Jul 2026 — Present

Temporary Delivery Driver

Viktor Nußbaumer Bestes für Küche und Gastlichkeit GmbH & Co. KG · Kürnach, Germany

Jul 2024 — Aug 2024

Picker

Hans Geis GmbH + Co KG · Kürnach, Germany

Jul 2022 — Aug 2022

Projects

Raven's Progressive Matrices — Abstract Visual Reasoning

Python · PyTorch

  • Team project attempting to replicate "An Interpretable Neuro-symbolic Model for Raven's Progressive Matrices Reasoning" (Zhao et al., 2023).
  • Reproduced two published dataset generators, RAVEN and RAVEN-FAIR.
  • Reproduced the paper's three-stage neuro-symbolic pipeline — CNN → supervised VAE for semantic attribute extraction → cognitive-map relational reasoning — reaching 94% prediction accuracy.

Mar 2026 — Aug 2026 · Cognitive Systems, Universität Bamberg

LLM-Powered GDPR Compliance Classifier

Python · SvelteKit · PostgreSQL · Docker · Ollama

  • Built a tool that classifies survey questions against the special categories of personal data defined in Art. 9 GDPR.
  • Developed the full stack: SvelteKit/TypeScript frontend and API routes, PostgreSQL schema, and a three-container Docker Compose deployment. Kubernetes load balancing is planned.
  • Ran all LLM inference locally through Ollama so sensitive institute data never leaves the network.

Oct 2025 — Present · LIfBi

Both projects are closed source.

Education

TU Dublin

ERASMUS Exchange (Computing) · Dublin, Ireland

Sep 2026 — May 2027

Universität Bamberg

B.Sc. Angewandte Informatik (Applied Computer Science) · Bamberg, Germany

  • Average score 1.98 across 99 credited ECTS.

Oct 2024 — Present

Röntgen-Gymnasium Würzburg

Abitur, grade 2.0 · Würzburg, Germany

Sep 2016 — Jul 2024

Grades

Average score 1.98 · 99 credited ECTS

Discrete Modelling
1.3
Java Programming
1.3
Linear Algebra
1.3
DevOps for AI Systems
1.3
Analysis
1.7
Introduction to AI
1.7
Introduction to ML
1.7
Statistics & Data Science
2.0
European Data Protection Law
2.3

Skills

Languages
PythonJavaTypeScriptRSPSS (introductory)C-style languages (introductory)
Libraries & Tools
PyTorchNumPyPandasDockerOllamaUMLGit
Concepts
Machine LearningData ScienceDistributed SystemsObject-Oriented Architecture Patterns
Practices
Agile Project ManagementRequirements EngineeringContinuous IntegrationSoftware Maintenance
Spoken
German (native)English (certified B2)