BIFOLD - Berlin Institute for the Foundations of Learning and Data

Too hot for the tracks. Too real to ignore. Leipzig, June 2026: melted joint compound flooded rails and switches, trams stopped entirely. This isn't a one-off. It's what happens when infrastructure built for yesterday's climate meets today's reality. For Batch III of Berliner Startup Stipendium Science & Startups is looking for founding teams working on exactly this: strengthening the resilience of critical infrastructure, whether that's energy, water, mobility, digital systems or AI-driven monitoring and prediction. Selected teams get 12 months of public, equity-free funding, backed by Berlin's university network. If you're building in this space, or know someone who is, don't miss the last obligatory consultation slot on July 31, applications close August 3, 12 PM.

📢 We're hiring! Our group at BIFOLD - Berlin Institute for the Foundations of Learning and Data is looking for a highly motivated Postdoctoral Researcher with a strong background in Agentic AI, data management, and machine learning. 🌍The successful candidate will conduct high-impact research on AI agents for Earth Observation (EO) that can interact with users to support environmental and climate intelligence. Research topics include, but are not limited to: 🔹Multi-agent systems 🔹Hallucination detection and mitigation 🔹Post-training methods for AI models 🔹Self-evolving and efficient AI systems 🔹Evaluation and benchmarking frameworks This is a 3-year fully funded postdoctoral position as part of the Horizon Europe project PROTEUS – The Copernicus Agent for Next-Generation Environmental Intelligence. 📅 If you are interested, please send your application to demir@bigearth.eu by August 15.

🧡 Research Spotlights from our Community   What are other AI Grid members in your field currently working on? This Thursday, our next exclusive AI Grid member event gives you a glimpse into ongoing research from within the community, with three short talks on current challenges in AI. The spotlight topics:   🧠 What LLMs Cannot Know: Making implicit knowledge explicit   ⚖️ How Language Models Value Arguments: Understanding different arguments in a debate   🕸️ Decomposing Ontology Engineering: An agentic pipeline for conceptualization   If you are interested in LLMs, reasoning, NLP, knowledge representation or agentic AI, join us to hear what your peers are working on, exchange ideas and discuss future directions with fellow researchers.   📅 30 July 2026 🕑 2:00–3:30 p.m. 💻 Online via Teams AI Grid members can find the registration link in their email or on Slack.   Formats like this are what AI Grid is all about: creating space for young AI researchers to share their work, learn from each other and connect across fields.   🚀 Not part of AI Grid yet?   If you are a Master’s or PhD student at a German university or university of applied sciences working on AI-related topics, AI Grid connects you with peers, mentors and exclusive formats like this one.   🔗 Learn more and apply: https://lnkd.in/e65948U4   #AIGrid #AIResearch #AICommunity #LLMs #NLP #KnowledgeRepresentation #AgenticAI #ResearchToImpact   Stefan Jazdzejewski, Ute Schmid, Johannes Winter, Jack Thoms, Mira Mezini, Laure Poirson, Sven Trendow, Alina Zaslavska, Varenia-Louisa Weiland

🎙️ Ask Me Anything – How do I get into AI for Cybersecurity? Hey everyone, We're excited to invite you to our 4th edition of "Ask me Anything". Our next guest is Linus Folkerts, a former member of Bliss, who now specialises on Cyber capability testing at the AI Security Institute (AISI). Having studied Cybersecurity as a Bachelor program already at Hochschule Aalen, which he expanded on in his Masters at UCL London, Linus will certainly have plenty to share about how you can make it in the field. This week's edition is for anyone interested in making a living from Cybersecurity and wondering about the current hot topics and companies. We would be happy to see you tomorrow, 29th of July 2026 📅 at 5PM 🕐 on our Discord Server. You can find the event at: https://lnkd.in/ekSzNCZF We will be on our ❓│ask-me-anything! speaking channel. You can drop your questions in advance in our ❓│ask-me-anything channel or bring them live on the day! See you at our discord event! 🚀 — Your Bliss Team

🔦 Past Paper Highlight: PRISM, interpreting polysemantic model features 📄 Capturing polysemanticity with PRISM: A multi-concept feature description framework 🔗 https://lnkd.in/dXXPthB4 💻 https://lnkd.in/dm-YR7Pp Most automated interpretability methods assume that individual neurons encode a single concept. The paper 'Capturing Polysemanticity with PRISM: A Multi-Concept Feature Description Framework' addresses this limitation by introducing a framework that generates multi-concept descriptions for LLM features. It introduces (i) a method for clustering high-activation inputs and labeling each cluster with an LLM to capture multiple concepts per neuron, (ii) a polysemanticity score to quantify how semantically diverse a neuron's associated concepts are, and (iii) a description score to evaluate how faithfully each concept label aligns with the neuron's activation distribution. Great work done by Laura Kopf, Nils Feldhus, Kirill Bykov, Philine Lou Bommer, Anna Hedström, Marina Marie-Claire Höhne, Oliver Eberle

🏹 PhD Position in Geometric Deep Learning and Graph Signal Processing 📍 Palaiseau 🇫🇷 ⏰ 6 September 🔗 https://bit.ly/4b7Ikgk

Keeping in close touch with academia is in the DNA of Databricks in general and Databricks in #Berlin in particular. We’re blessed with world-class groups in Data+AI in the larger Berlin eco-system and so it was great to (finally!) get together with the HPI Database group for a nice dinner. Our ties to Hasso Plattner Institute are close. HPI alumni are among the largest co-horts in the Berlin Databricks R&D office for interns and full-time employees alike. Bricksters regularly visit the HPI campus and once a year, a group of HPI students visit our office. And so it was great to spend time with HPI Profs and their visitors also outside the conferences – we missed you, Felix Naumann and we’ll definitely repeat this!  More on the Berlin data+AI ecosystem to come later in the year, also with our partners from TU Berlin and BIFOLD - Berlin Institute for the Foundations of Learning and Data. Stay tuned. Eleni Tzirita Zacharatou Tilmann Rabl Matthias Weidlich Boris Glavic Bernhard Seeger Marco Kroll Johan Lasperas Sebastian Hillig

#Hiring – Wir suchen eine:n neue:n AI Driven Data Manager*in - Beschäftigte*r mit abgeschl. wiss. Hochschulbildung - 90 % Arbeitszeit in Berlin. Bewerben Sie sich noch heute oder teilen Sie diese Stellenanzeige mit Ihrem Netzwerk.