Techday hackathon with NVIDIA

Last Friday at our Techday, Antonio Rueda Toicen and Stefan Wenz from NVIDIA visited our headquarters in Unterföhring and hosted a maker session with our team.

The focus was on the Nemotron model family: specifically Nemotron Parse, Embedding, Omni, and Rerank. Inspired by the NVIDIA NeMo-Retriever, we used Nemotron models to build a retrieval-augmented generation (RAG) ingestion pipeline for documents that can parse text, tables, charts, and infographics.

Nemotron is no stranger to us: recently, we packaged Nemotron 3.5 Lightning as GGUF for local use on a single 24 GB GPU, and taught the open-weights model to see using a small vision adapter.

Thanks to Antonio and Stefan for an exciting day and hands-on insights into the Nemotron model family!