Jan Philip Göpfert
AI Engineer and Founding Technical Lead
PhD in machine learning. I build LLM systems with coding agents and run hands-on AI workshops.
About Me
I lead the technical side of an early-stage startup that produces companies' CSRD sustainability reports, and I build almost everything with coding agents. Before that I was principal machine learning engineer at a computer vision startup.
My PhD at Bielefeld University was on adversarial robustness. I lectured there from 2015 to 2022, and I run hands-on AI workshops in industry and academia.
Through GAIA (Göpfert AI Applications) I take on selected consulting and workshops.
Areas of Expertise
LLM Pipelines
Pipelines from document intake to finished report, in which every extracted claim traces back to its place in the source.
Data Engineering
Typed data flow through ML and LLM code, with Pydantic schemas, Parquet and Postgres.
Model Evaluation
Evaluations for choosing models and providers that plant confounders and degrade the documents, and show where on the page a value was found or missed.
AI System Architecture
ML and LLM systems in production: several model providers with failover, tracing on every call, data held in the EU.
Teaching & Workshop Facilitation
Lectures and tutorials at Bielefeld University from 2015 to 2022, and hands-on AI workshops for industry and academia.
Machine Learning Research
Adversarial robustness, uncertainty quantification and human-machine interaction. Papers at ICANN, ESANN and IDA.
Project Highlights
Healthcare-Tech Review
Security & architecture review for a healthcare-tech startup
AI Training
AI literacy training for business and public administration
E-commerce Search
Multimodal semantic search for e-commerce
Model Training Speedup
Training pipeline rewrite, 200x faster
Community Engagement
Community engagement assistants for content creators
LLMs for Pharmacology
AI applications in pharmacology research
E-commerce Recommendations
Visual-based recommendations for fashion e-commerce
Lead Conversion Prediction
Predict lead conversion with customer journey analysis
Montessori Malvorlagen
Digitization Pipeline
Automated digitization pipeline for scanned documents
Selected Publications
Robustness in Machine Learning: Adversarial Perturbations, Explanations & Intuition
Jan Philip Göpfert
2022 | Bielefeld University
PhD thesis on adversarial robustness, explanations, and intuition in machine learning.
Deep Learning for Understanding Satellite Imagery: An Experimental Survey
Sharada Prasanna Mohanty, Jakub Czakon, Kamil A. Kaczmarek, ..., Jan Philip Göpfert, …
2020 | Frontiers in Artificial Intelligence
We explore automated satellite image analysis using deep learning and present five approaches based on U-Net and Mask R-CNN models, achieving impressive results in building detection.
Intuitiveness in Active Teaching
Jan Philip Göpfert, Ulrike Kuhl, Lukas Hindemith
2020 | IEEE Transactions on Human-Machine Systems
We propose intuitiveness as a property of machine learning algorithms, largely impacting how easy it is for users to interact with a given algorithm without any explicit instruction or training.