Volker Hartmann

• Performed groundwork on conversational AI by evaluating frameworks and possible research directions. • Spearheaded the development of an open-domain chatbot and its integration into a messenger app. • Co-developed a modular LLM-based long-range conversational AI with efficient LLM inference support (https://github.com/krafton-ai/MPC). • Worked on efficient LLM fine-tuning using low-rank approaches such as LoRA.

• Pre-trained and fine-tuned language models, with a focus on question-answering and text-summarization, using TPU and GPU. • Secured 1st place in Microsoft’s GLGE competition with CTRLgen submission using keyword-guided generation. • Accelerated language model inference by up to 4 times using quantization, mixed precision, knowledge distillation, and TF-XLA. • Implemented internal and external demos with streamlit, HTML, and JS for proprietary models.

Nutzerberatung im Bereich Mobilfunk

Abschlussarbeit im Bereich der graphischen Datenverarbeitung (Farbkalibrierung einer Industriekamera)

Vertiefung: Machine Learning, Natural Language Processing; Studienaustausch @SNU Seoul, Suedkorea (WS/SS 2017/2018)

Studienaustausch @TU Dublin (WS 2014)