Portrait of Lukas Laskowski

Senior Research Engineer @ SAP Business AI. I build AI systems that turn complex enterprise data into usable knowledge—from learned semantic models to production applications.

RESEARCH OUTPUTS · AI × enterprise data
Systems · benchmarks · learning methods
Ontology learning · data integration · knowledge graphs
Lead-author work at VLDB · SIGMOD · EDBT
--:--:-- BER · Berlin, Germany
SCROLL
About

AI that understands enterprise data

My work sits at the intersection of AI, data systems, and knowledge representation. I develop systems that discover structure and meaning in complex enterprise data by combining LLMs, representation learning, human feedback, and symbolic methods. The result is reliable context for AI: semantic data layers, ontologies, and knowledge graphs learned from data rather than modelled entirely by hand.

I completed my PhD at the Hasso Plattner Institute in close cooperation with SAP SE, researching AI methods for ontology learning and data integration, with lead-author work at VLDB and SIGMOD. At SAP Business AI, I now translate this research and broader applied-AI experience into enterprise-scale systems.

My practical experience extends beyond knowledge graphs: industry engineering since 2018, from building a global Industrial IoT platform at Siemens Energy to applying computer vision to wildlife conservation in Rwanda.

Experience

AI research, engineered for practice

2026 — now

Senior Research Engineer

Developing AI systems for enterprise data, with a focus on semantic grounding, ontology learning, and knowledge graphs—and translating research into scalable product capabilities.

11/2023 — 04/2026

Doctoral Researcher

Research at the intersection of AI, databases, and knowledge representation, developing learning-based methods for ontology construction and data integration in close cooperation with SAP SE. Lead-author publications at VLDB and SIGMOD.

Feb — May 2025

Visiting Researcher

Human-in-the-loop ontology learning with Prof. Padhraic Smyth, combining databases, Semantic Web and AI methods.

2024 — 2025

Technical Advisor & AI Engineer

Self-employed, partnering with Tehanu

Consulting on wildlife re-identification AI for a species-conservation startup based in Rwanda.

2018 — 2023

Software Engineer (Working Student)

Primary owner of the architecture and development of the automated edge-provisioning solution for the global "Connected Factory" IIoT platform, benefiting 80+ factories; full-stack applications on AWS; technical management of external staff in Europe and India. Recognised as a successful digitalisation project.

2021

Cloud Operations Engineer (Internship)

SAP SE, Master Data Governance Cloud

Service administration on the SAP Cloud platform and automated deployment pipelines.

Ph.D. HPI, 2026|M.Sc. Data Engineering, HPI — 1.2, with distinction · Thesis: NumbER|B.Sc. IT-Systems Engineering, HPI — 1.8
Teaching: 5 seminars on AI and data cleaning, 2 master projects on data integration, and supervision of 2 master theses (multimodal RAG, 2025; foundation models for flight data, 2025/26).
Publications

Selected publications

Ongoingin progresslead author

Hamilton: A Human-in-the-Loop Ontology Learning System for Relational Databases

Bringing expert feedback into the ontology-learning loop, building on the research visit at UC Irvine.

VLDB2026 · A*lead author

Schuyler: Self-Supervised Clustering of Tables in Relational Databases

Clusters database tables by combining structural and semantic signals via triplet-loss fine-tuning of an LLM, with state-of-the-art results (+0.13 ARI, +0.10 AMI) on a new five-database benchmark.

SIGMOD2026 · A*lead author

Burr: A Benchmark for Ontology Learning from Relational Databases

54 real-world and synthetic scenarios plus a novel mapping-based metric, showing that rule-based methods still lead while LLM-based approaches hold substantial promise.

NeurIPS2025 · Spotlight

Learning Conditional Marked Event Sequences with Mixed Data Types

An intensity-free, Transformer-based marked temporal point process that jointly models event times and mixed-type marks with normalizing flows. With UC Irvine; Spotlight, top 3% of submissions.

EDBT2025 · Asupervisor

Prisma: A Privacy-Preserving Schema Matcher using Functional Dependencies

Matches columns across schemas using functional dependencies and graph embeddings instead of names or values, outperforming existing methods on multi-table databases with divergent or encrypted encodings.

VLDB2022 · A*lead author

Frost: A Platform for Benchmarking and Exploring Data Matching Results

The first platform to evaluate deduplication solutions on quality, cost and effort together, with interactive result exploration that goes beyond accuracy-only benchmarks.

WorkshopCamera Traps, AI & Ecology

GorillaVision: Open-Set Re-Identification of Wild Gorillas

YOLOv7 face detection plus a ViT embedding model trained with triplet loss and k-NN classification: reliable identification of unseen individuals at 60–90% accuracy.

Highlights

Along the way

Dec 2025

NeurIPS Spotlight (~top 3%)

Our MTPP paper with UC Irvine was selected as a Spotlight.

Invited talk

Talk at MIT

On entity resolution on numerical data and ontology engineering.

2025

Research visit, UC Irvine

Human-in-the-loop ontology learning with Prof. Padhraic Smyth.

2022

Paper talk at VLDB, Sydney

Presented Frost at the 48th VLDB conference.

2024

Young Volunteer of the Year, Berlin

For board work as vice chair of the tennis department of VfB Hermsdorf (250+ members).

Talk

SAP Signavio Knowledge Club

On integrating enterprise data into a unified knowledge graph.

Also: Graduate Award of the German Physical Society (2018) · Co-founder of the HPI Entrepreneur Club (80+ members).
Contact

Let's talk

Enterprise AI, machine learning for data systems, ontology learning, knowledge graphs, or AI for conservation? I'm happy to hear from you.

lukas.laskowski@gmx.de