Brussels / 1 & 2 February 2025

schedule

Low-level AI Engineering and Hacking

Room: UB2.252A (Lameere)
Calendar: iCal, xCal


09 10 11 12 13 14 15 16 17 18
Sunday Hugging Face ecosystem for Local AI/ ML
 The Local AI Rebellion
 ZML: A High-Performance AI Inference Stack Built for Production and Multi-Accelerator Deployment
 History and advances of quantization in llama.cpp
 quantizing your GGUF models using iterative refinement of the importance matrix
 Apache Arrow: The Great Library Unifier
 Bringing AI to Wearable Systems: Integrating Vision, Audio, and Sensors on Constrained Hardware
 How Llamagator helps to implement LLM-as-a-Judge concept on your local machine
 Compositional LLMs for Assisted competitive coding
 The Model Openness Framework (MOF)
 Building AI Applications on Kubernetes: Leveraging Instructlab and the Bee Agent Framework
 GPUStack: Building a Simple and Scalable Management Experience for Diverse AI Models
 Self-hosted LLMs at a scale with Paddler
 RamaLama: Making working with AI Models Boring
 Building AI Applications from your desktop with Podman AI Lab
 From Supercomputer to Raspberry Pi: Building Open Source Polish Language Models
 Tricks Learned from Training Large Open-Source Models
 Synthetic Data: The Secret Ingredient in Better Language Models
 LLM Tool use in vLLM
 Scoping out the Tenstorrent Wormhole
 Building a new GGML backend: How, Challenges and Oppertunities with Novel Accelerators
 Porting GGML to the NUX Kernel Development Framework.
 Accelerating AI with open source hardware and software
 The bare metal perspective on AMD's GPU ASICs
 wllama: bringing llama.cpp to the web
 Milliwatt sized Machine Learning on microcontrollers with emlearn

Read the Call for Papers at https://aifoundry.org/fosdem-2025-low-level-ai-engineering-hacking-dev-room.

Event Speakers Start End

Sunday
  Hugging Face ecosystem for Local AI/ ML
 VB 09:05 09:30
  The Local AI Rebellion
 Justine Tunney 09:30 10:00
  ZML: A High-Performance AI Inference Stack Built for Production and Multi-Accelerator Deployment
 Rene Schallner, Guillaume Wenzek 10:00 10:30
  History and advances of quantization in llama.cpp
 Tanya Dadasheva, Iwan Kawrakow 10:30 11:00
  quantizing your GGUF models using iterative refinement of the importance matrix
 Robert Collins 11:00 11:20
  Apache Arrow: The Great Library Unifier
 Matthew Topol 11:20 11:50
  Bringing AI to Wearable Systems: Integrating Vision, Audio, and Sensors on Constrained Hardware
 Kris Kersey 11:50 11:55
  How Llamagator helps to implement LLM-as-a-Judge concept on your local machine
 Sergy Sergyenko 11:55 12:00
  Compositional LLMs for Assisted competitive coding
 Ritwik Agarwal 12:00 12:05
  The Model Openness Framework (MOF)
 Arnaud Le Hors 12:05 12:10
  Building AI Applications on Kubernetes: Leveraging Instructlab and the Bee Agent Framework
 Martin Hickey, Paul Schweigert 12:10 12:15
  GPUStack: Building a Simple and Scalable Management Experience for Diverse AI Models
 Lawrence Li 12:20 12:40
  Self-hosted LLMs at a scale with Paddler
 Mateusz Charytoniuk 12:40 13:00
  RamaLama: Making working with AI Models Boring
 Eric Curtin 13:00 13:20
  Building AI Applications from your desktop with Podman AI Lab
 Cedric Clyburn, Stevan Le Meur 13:20 13:40
  From Supercomputer to Raspberry Pi: Building Open Source Polish Language Models
 Bielik Team, Maciej, Pawel Cyrta, Adrian 13:40 13:55
  Tricks Learned from Training Large Open-Source Models
 Marcus Edel 13:55 14:10
  Synthetic Data: The Secret Ingredient in Better Language Models
 Carol Chen, Cedric Clyburn 14:10 14:25
  LLM Tool use in vLLM
 Max de Bayser 14:25 14:40
  Scoping out the Tenstorrent Wormhole
 Peter Cawley 14:40 15:00
  Building a new GGML backend: How, Challenges and Oppertunities with Novel Accelerators
 Martin Chang 15:00 15:20
  Porting GGML to the NUX Kernel Development Framework.
 Gianluca Guida 15:20 15:40
  Accelerating AI with open source hardware and software
 William Jones 15:40 16:00
  The bare metal perspective on AMD's GPU ASICs
 Jon Chesterfield 16:00 16:20
  wllama: bringing llama.cpp to the web
 Xuan-Son Nguyen 16:20 16:40
  Milliwatt sized Machine Learning on microcontrollers with emlearn
 Jon Nordby 16:40 17:00