• Hauptseite
  • Writings
  • Workshops
  • 👾Über mich
    • Allgemeines
    • Neurodivergenz
    • Musik
    • CV
  • Kontakt & Impressum

Eri Schölzel (they/them)

Senior AI Developer / Machine Learning Engineer

Contact

contact@luux.dev

luux.dev
Turnerweg 2
01445 Radebeul
Germany

Tech Stack

Main Focus
Python Data Pipelines Applied AI Deep Learning
Infrastructure & Tooling
Containerization CI/CD Service APIs Git Data Version Control (DVC)
Hosting & Administration
Nextcloud Plesk Wordpress Grav
Used long ago, but rusty
LaTeX Java C#

Languages

German Native
English Fluent
Japanese Basic

Professional Software Engineering

09/2023 – Present

Machine Learning Engineer (Senior Contributor)

elevait GmbH & Co. KG
Feature Team Information Extraction
Implementing 0-shot extraction and automated prompt-tuning using dspy, integrated into company-wide language model training workflows and reducing manual data handling. Maturing to an extraction platform used across the company.
Azure OpenAI • dspy • LLM • Small language models
ai:cm
Led development of an LLM-based classification and information extraction pipeline. Hierarchical classification system with 100+ classes from the construction domain. RAG and benchmark pipeline for information retrieval using datasets with 10k++ samples, used across multiple internal teams.
LangChain • RAG • Milvus • Pydantic • Azure OpenAI • OCR
02/2021 – 09/2023

Machine Learning Engineer (Junior Contributor)

elevait GmbH & Co. KG (former AI4BD Deutschland GmbH)
in:plans
Analysis and processing of large-scale construction plans using computer vision and geometry analysis, including classification, georeferencing and key/value pair extraction.
DVC • PyMuPDF • Pdf geometry • Tensorflow • OpenCV • Scikit-learn
OCR for order processing of medical devices.
Custom OCR model and training implementation for processing order forms.
TensorFlow • Data Versioning Control (DVC) • Keras
12/2018 – 10/2020

Working Student

T-Systems Multimedia Solutions Dresden
AI4Test
Developed object detection models for GUI element detection using weak supervision and transfer learning in the domain of test automation. Designed annotation principles and instructed a team of annotators, including data QA.
Object Detection • Weak Supervision • Transfer Learning • Darknet (ML Library) • Selenium • Web crawling
10/2016 – 03/2018

Student Assistant

Competence Center for Scalable Data Services and Solutions (ScaDS) Dresden
OCR on Messtischblättern
OCR on old German topographical maps using Deep Learning.
OCR • Keras • Caffe (Deep Learning Framework) • Slurm (HPC) • Computer Vision

Formal Education

10/2014 – 10/2020

Diploma Computer Science (1.5)

Technical University Dresden

Minor Subject: Acoustics (2015-2018).

Diploma Thesis (1.0): Evaluation of Weak Supervision and Transfer Learning for GUI Element Detection (Cooperation with T-Systems MMS).

Minor Thesis (1.3): Classification of motion patterns (Cooperation with Evomo).

09/2006 – 07/2014

Abitur (1.6)

Christian-Weise-Gymnasium Zittau

Engagement & Community

Neurodiversity Advocacy

Workshops on intersection of neurodivergence, communication and consent (smjg e.V. and more)

Speaker at community events and conferences (e.g. Queerpride, LAG Mädchen* Sachsen)

Community building and support (e.g. Neuroqueer Community Dresden, 38c3/39c3, Neurodiversity Community of Practice @ elevait)

Rope ('Shibari'/'Kinbaku')

Regularly participating at international events, workshops and festivals since 2022

Occasionally organizing workshops with high focus on communication, consent, risk-awareness and inclusion, e.g. at DeDeCo e.V.

Organizing queer-centered rope jams in Dresden, see boundaries.fun.

Music

Many years of choir singing experience, including 10 years in Landesjugendchor Sachsen. Concert tours in 8 countries and multiple CD recordings.

Proficiency in overtone singing

Grav was with by Trilby Media.

  • Hauptseite
  • Writings
  • Workshops
  • 👾Über mich
    • Allgemeines
    • Neurodivergenz
    • Musik
    • CV
  • Kontakt & Impressum