BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Mondai - ECPv6.15.12.2//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:Mondai
X-ORIGINAL-URL:https://mondai.tudelftcampus.nl
X-WR-CALDESC:Evenementen voor Mondai
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:Europe/Amsterdam
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:20250330T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:20251026T010000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:20260329T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:20261025T010000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:20270328T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:20271031T010000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=Europe/Amsterdam:20260316T100000
DTEND;TZID=Europe/Amsterdam:20260316T173000
DTSTAMP:20260407T180410
CREATED:20260129T095419Z
LAST-MODIFIED:20260304T134839Z
UID:10636-1773655200-1773682200@mondai.tudelftcampus.nl
SUMMARY:Towards Democratized Academic LLM Hosting
DESCRIPTION:Towards Democratized Academic LLM Hosting\nA 4TU Joint Workshop on Local Inference Infrastructure (hosted at TU Delft) Large Language Models (LLMs) have become core research infrastructure for academia\, underpinning research\, education\, and innovation across disciplines. Across Dutch and European academia\, multiple initiatives are currently underway to host\, operate\, and govern LLMs locally\, driven by concerns around data sovereignty\, sustainability\, cost\, transparency\, and responsible use. \nThis 4TU.NIRICT-funded workshop brings together currently siloed academic hosting initiatives across Dutch institutions (eg\, Delft\, TU/e\, VU\, WUR\, SURF) to foster knowledge exchange\, technical alignment\, and collaboration around LLM inference infrastructure.  \nThe workshop addresses LLM hosting as a socio-technical challenge\, spanning infrastructure and deployment models\, energy efficiency\, governance and operational sustainability\, tooling ecosystems\, and integration into research and education. It lays the groundwork for a coordinated\, academic-first Dutch ecosystem for responsible\, open LLM hosting. \nFull Programme March 16 10.00 – 10.30 Welcome and Registration \n10.30 – 10.35 Opening Remarks \n10:35-11:20 Keynote 1 - Alexandre Strube – Lessons from the Blablador project\nIn this talk\, we will show some of the challenges and lessons learned while hosting an LLM inference service. BLABLADOR has been running as a single-man-operation for most of its existence\, until very recently. \nThis talk is more of a conversation\, where we will explore themes such as: \n\nReliability\nSoftware Ecosystem and evolution\nFrom small clusters to supercomputers to cloud: quite a messy journey\nWho can use it and handling user abuse\nAdvertising the service\nChoosing which models to offer\nPolitical consequences of model choice\nHow to convince management that it’s worth it\nPresent and future of the service\nJAIF and JARVIS: An AI factory for Blablador\nQuestions from the audience\n\nDr. Alexandre Strube \n\nMaintainer of LMod\, the Supercomputers’ module system\nMaintainer of FastChat\nUbuntu Member\, Debian Developer\n\nDr. Alexandre Strube has been employed at the Juelich Supercomputing Centre for approximately 15 years. During this time\, he has engaged in a diverse range of activities\, encompassing raw performance modeling\, supercomputer benchmarking\, and teaching\, among many others. Notably\, he has been a member of the Helmholtz AI consultants team since 2019\, where he has contributed to the democratization of artificial intelligence within academic institutions. \nDr. Strube has been responsible for the development\, hosting\, and maintenance of BLABLADOR\, the LLM inference infrastructure for the Helmholtz Foundation and German Academia\, for approximately four years. \n11.20 – 11.35 Coffee Break \n11:35-12:35 Breakout Sessions (Round 1)\nMorning (Round 1) \n\nEnergy consumption: Benchmarking approaches; power-aware optimization strategies\nGovernance & Operational Sustainability: Service creation pathways; Financing and maintenance models; Security\, privacy\, and access control; Architectural quality assurance\nTooling around model hosting: Monitoring and observability; guardrails and safety layers; RAG and MCPs; Service abstractions to end-users\n\n12.35 – 13.35 Lunch \n13:35-14:20 Keynote 2 - Julio Alexandrino de Oliveira Filho – National perspectives and GPT-NL\nGPT-NL is a national   initiative to build a large language model grounded in the Dutch language\, culture\, and public values from scratch. Beyond delivering a performant model\, the GPT-NL project set out to explore what it means in practice to develop a sovereign language model—one that is transparent\, inclusive\, compliant with Dutch and European law\, and embedded in a broader public-interest ecosystem. \nThis keynote reflects on the practical experiences accumulated throughout the GPT-NL project in building such a model while simultaneously fostering a diverse and sustainable LLM community. Drawing on lessons learned during data acquisition\, curation\, and training\, the talk will discuss legal and regulatory considerations\, approaches to fair and lawful data use\, and the organizational challenges of aligning academic\, public\, and commercial stakeholders. We give special attention to infrastructure-related questions\, including large-scale training\, hosting\, customization\, and the role of shared HPC and hosting facilities in enabling broader participation. \nBeyond the model itself\, the keynote will examine how the GPT-NL project has sought for community building across researchers\, LLM and AI developers\, data providers\, legal experts\, infrastructure operators\, and application developers—both nationally and in a broader European context. The talk will outline envisioned application domains\, strategies for responsible dissemination\, and open research directions that arise when a language model is treated as a shared societal asset rather than a closed product. \nBy sharing successes\, open challenges\, and unresolved questions\, this keynote aims to contribute concrete insights to the discussion on democratized academic LLM community and to inform future national and European efforts toward sovereign\, community-centered language models. \nDr. Rer Nat. Julio A. de Oliveira Filho researches and develops the next generation of intelligent autonomous systems—spanning robotics\, quantum networks\, IoT sensor systems\, automotive technologies\, and modern Defense applications.  \nAs the leading architect of GPT-NL\, the first Dutch-English large language model trained fully from scratch\, and one of the creators of NetSquid\, the world’s most widely used quantum network simulator\,  Julio thrives at the intersection of advanced AI\, complex systems\, and cutting-edge engineering.  \nHe is passionate about elevating AI engineering into a mature discipline—where autonomous systems are not only powerful\, but reliable\, safe\, and explainable by design. \n14.20 – 14.25 Room Change \n14:25-15:25 Breakout Sessions (Round 2)\nAfternoon(Round 2) \n\nAccelerate production-ready LLM deployment using commercial solutions: Why is this relevant: Possible use cases; NVIDIA AI Enterprise; Microsoft AI Foundry\nEducation & Research Use Cases: Practical implementations; researcher-education feedback loops; implications for curricula; Policy & ethics considerations\nInfrastructure & Deployment: Utilization strategies; deployment models: on-prem\, shared\, and hybrid\n\n15.25 – 15.40 Coffee Break \n15:40-16:40 Panel Discussion: Future of AI & LLM/model hosting in the Netherlands and Europe\nA moderated discussion with academic leaders and national stakeholders exploring how academic institutions should position themselves in a rapidly evolving landscape shaped by national and European initiatives\, open and proprietary models\, emerging AI hubs\, and increasing policy attention. Particular emphasis will be placed on the quality\, governance\, and sustainability of LLM services in an academic context. \n16.40 – 17.40 Borrel & Networking \nSpeakers Alexandre Strube (Helmholtz AI) - Keynote 1\nIn this talk\, we will show some of the challenges and lessons learned while hosting an LLM inference service. BLABLADOR has been running as a single-man-operation for most of its existence\, until very recently. \nThis talk is more of a conversation\, where we will explore themes such as: \n\nReliability\nSoftware Ecosystem and evolution\nFrom small clusters to supercomputers to cloud: quite a messy journey\nWho can use it and handling user abuse\nAdvertising the service\nChoosing which models to offer\nPolitical consequences of model choice\nHow to convince management that it’s worth it\nPresent and future of the service\nJAIF and JARVIS: An AI factory for Blablador\nQuestions from the audience\n\nDr. Alexandre Strube \n\nMaintainer of LMod\, the Supercomputers’ module system\nMaintainer of FastChat\nUbuntu Member\, Debian Developer\n\nDr. Alexandre Strube has been employed at the Juelich Supercomputing Centre for approximately 15 years. During this time\, he has engaged in a diverse range of activities\, encompassing raw performance modeling\, supercomputer benchmarking\, and teaching\, among many others. Notably\, he has been a member of the Helmholtz AI consultants team since 2019\, where he has contributed to the democratization of artificial intelligence within academic institutions. \nDr. Strube has been responsible for the development\, hosting\, and maintenance of BLABLADOR\, the LLM inference infrastructure for the Helmholtz Foundation and German Academia\, for approximately four years. \nJulio Alexandrino de Oliveira Filho (TNO) - Keynote 2\nDr. Rer Nat. Julio A. de Oliveira Filho researches and develops the next generation of intelligent autonomous systems—spanning robotics\, quantum networks\, IoT sensor systems\, automotive technologies\, and modern Defense applications.  \nAs the leading architect of GPT-NL\, the first Dutch-English large language model trained fully from scratch\, and one of the creators of NetSquid\, the world’s most widely used quantum network simulator\,  Julio thrives at the intersection of advanced AI\, complex systems\, and cutting-edge engineering.  \nHe is passionate about elevating AI engineering into a mature discipline—where autonomous systems are not only powerful\, but reliable\, safe\, and explainable by design. \nSovereign Language Models in Practice: The GPT-NL Endeavour in LLM Community Building \nGPT-NL is a national   initiative to build a large language model grounded in the Dutch language\, culture\, and public values from scratch. Beyond delivering a performant model\, the GPT-NL project set out to explore what it means in practice to develop a sovereign language model—one that is transparent\, inclusive\, compliant with Dutch and European law\, and embedded in a broader public-interest ecosystem. \nThis keynote reflects on the practical experiences accumulated throughout the GPT-NL project in building such a model while simultaneously fostering a diverse and sustainable LLM community. Drawing on lessons learned during data acquisition\, curation\, and training\, the talk will discuss legal and regulatory considerations\, approaches to fair and lawful data use\, and the organizational challenges of aligning academic\, public\, and commercial stakeholders. We give special attention to infrastructure-related questions\, including large-scale training\, hosting\, customization\, and the role of shared HPC and hosting facilities in enabling broader participation. \nBeyond the model itself\, the keynote will examine how the GPT-NL project has sought for community building across researchers\, LLM and AI developers\, data providers\, legal experts\, infrastructure operators\, and application developers—both nationally and in a broader European context. The talk will outline envisioned application domains\, strategies for responsible dissemination\, and open research directions that arise when a language model is treated as a shared societal asset rather than a closed product. \nBy sharing successes\, open challenges\, and unresolved questions\, this keynote aims to contribute concrete insights to the discussion on democratized academic LLM community and to inform future national and European efforts toward sovereign\, community-centered language models. \nCorry Wouters (TU/e) - Panelist\nCorry Wouters is CIO and Director of Library & Information Services at Eindhoven University of Technology (TU/e). With over 30 years in commercial tech and healthcare\, she has led major digital transformations in EHRs\, secure interoperability\, and data governance\, delivering scalable\, secure services. At TU/e\, she drives digital innovation to support education\, research\, and societal impact. In 2025\, she was named one of the most innovative leaders in the Netherlands. \nLinkedIn \nArie van Deursen (TU Delft) - Panelist\nArie van Deursen is a professor in software engineering at the department of Software Technology in the faculty of Electrical Engineering\, Mathematics\, and Computer Science. He served as head of department from 2016-2023\, and presently chairs the Software Engineering Research Group (SERG). \nHe is scientific co-director of AI4SE\, a five year collaboration between JetBrains and TU Delft investigating novel use of artificial intelligence in software engineering. This lab hosts 10 PhD students\, dozens of MSc and Bsc students\, and many TU Delft faculty members and JetBrains specialists. \nHe is co-founder of two companies: The Software Improvement Group (2000) and PerfectXL (2010). \nDamian Podareanu (SURF) - Panelist\nLinkedIn \nNiels Taatgen (RUG) - Panelist\nRead all about Niels and his work on the infopage of RUG \nDr. Thijs van der Plas (WUR)\nThijs van der Plas is a postdoctoral AI researcher at Wageningen University & Research in the newly formed AI group. He currently works on ecological data integration in the Dutch LTER-LIFE project and on explainable AI in the ESA-funded AETHER project. Thijs has a background in physics at Radboud University\, holds a DPhil in computational neuroscience from the University of Oxford and worked as Research Associate at the Alan Turing Institute. He currently builds multimodal and explainable AI methods for monitoring biodiversity at scale\, as well as LLM systems for improving data management. \nLink additional info
URL:https://mondai.tudelftcampus.nl/event/towards-democratized-academic-llm-hosting/
LOCATION:Mondai | House of AI – Delft\, Molengraaffsingel 29\, Delft\, 2629 JD
ATTACH;FMTTYPE=image/png:https://mondai.tudelftcampus.nl/wp-content/uploads/sites/7/2022/05/DO_Mondai_LOGO_futuristischblauw.png
END:VEVENT
END:VCALENDAR