Luma·Builder event·IRL

UNCONFERENCE: Engineering AI Together #12

MON 23 Nov
Mon 23 Nov 2026, 18:00Europe/Berlin
Tomorrow Starts in 1d 0h
Merantix AI CampusMerantix AI Campus, Berlin, Germany
The details

About this one

Join us for the 12th edition of the Unconference: Engineering AI Together. A no-hierarchy gathering for software engineers, AI/ML practitioners, and technical builders to spend an evening exchanging ideas, pressure-testing approaches, and going deep on the problems they are actually working on. Unlike a traditional conference, there is no preset agenda and no speaker lineup. The sessions are proposed and shaped by the people in the room. Bring a technical question, a problem you are stuck on, something you have learned recently, or simply join a discussion that catches your attention.…

Read the full details on Luma ↗
The basics

Good to know

When
Mon 23 Nov 2026, 18:00
Where
Merantix AI CampusMerantix AI Campus, Berlin, Germany
Type
Builder event
Format
IRL
Get oriented

Where it’s going down

25 km10 km5 km
3.8 km from Berlin centreOpen in maps ↗
Keep the momentum

More in Berlin

See all Berlin events →
MON 23Nov
Mon 23 Nov 2026, 18:00 ESMT Berlin The Venture Club

Have a startup idea and maybe a co-founder? Some legal decisions are much easier to get right at the start than to fix later. The Venture Club is teaming up with V14, a Berlin law firm that advises startups and investors, for a short version of their Legal Starter Pack. It’s built for founders who are still shaping their idea and building their team. Marie Hillebrand and Simonié Schlombs (V14) will cover: IP: how to protect what you’re building, and make sure the company owns it rather than individual founders Co-founder agreements: equity splits, vesting, roles, and what happens if a co-founder leaves Starting to operate in Germany/Europe: the minimum you need to get going legally, from choosing a legal form to what comes before the notary The session ends with an open Q&A, so bring your questions. 📍 ESMT Berlin🗓 Monday 23 November · 18:00–19:00, followed by networking

29 going via Luma
Add to calendar
⋯
MON 23Nov
Mon 23 Nov 2026, 18:30 Berlin LSE Generate Events

AI is entering its deployment era... As AI models become increasingly powerful and accessible, the real challenge is making them work at scale. BCG estimates that 70% of successful AI transformation efforts depend on people, processes and organisational change - not technology. So what does it actually take to turn AI capabilities into measurable business impact? AI ambition meets business impact: join LSE Generate and BCG X in Berlin for an evening with Sina Stöhr, Partner in BCG’s Berlin office. BCG X is Boston Consulting Group’s tech build and design unit, bringing together more than 3,000 technologists, data scientists, designers and entrepreneurs. What we’ll explore Drawing on real-world examples of AI transformation, we’ll explore: Beyond the pilot: Why promising AI initiatives fail to scale. Rethinking workflows: Why AI adoption requires more than new tools. Delivering impact: How leading companies turn AI into measurable productivity gains. Expect practical insights, an interactive discussion and the opportunity to connect with Berlin’s entrepreneurial and tech community. We can't wait to see you there!

via Luma
Add to calendar
⋯
MON 23Nov
Mon 23 Nov 2026, 18:30 Merantix AI Campus BRISA

Today, we have a special one: Denis Shcherba presents his own paper, Estimate, Don’t Imitate: Reusing Differentiable State-Based Policies for Visuomotor Control (Shcherba et al., 2026), written right here in Berlin with Marc Toussaint (TU Berlin) and Wojciech Samek (Fraunhofer HHI). A rare chance to ask the author directly why things were done the way they were. In simulation a robot can cheat: it knows exactly where every object is. Policies trained with this privileged state learn contact-rich behaviour well, but a real robot only gets noisy camera images. The usual fix is teacher-student distillation, where a vision policy learns to copy the expert’s actions. That forces the student to do two jobs at once: work out the state from pixels, and relearn a mapping from state to action that the expert already has. This paper keeps the expert and only learns a visual state estimator. The catch is that an error in the estimate can matter a lot or not at all, depending on what the expert does with it. So the estimator is trained on two signals: plain state regression, and an action-consistency loss backpropagated through the frozen, differentiable expert. Across five goal-conditioned manipulation tasks, keeping the expert consistently beats direct pixel-to-action imitation trained on the same demonstrations. On a real Panda arm it reaches 76% success, without retraining the expert. No preparation needed. Come with questions for the author, disagreements or your own view on end-to-end versus modular. 🎟️ There are only a limited number of spots, so we can keep the session personal and interactive. Please register early and only attend if you’ve received a ticket. 📍 Hosted at the Merantix AI Campus on November 23rd. Doors open 18:30, we start at 18:45. 👋 New here? Join the community group - everything else gets announced there first.

13 going via Luma
Add to calendar
⋯
Same community

More from Merantix AI Campus

MON 23Nov
Mon 23 Nov 2026, 18:30 Merantix AI Campus BRISA Berlin

Today, we have a special one: Denis Shcherba presents his own paper, Estimate, Don’t Imitate: Reusing Differentiable State-Based Policies for Visuomotor Control (Shcherba et al., 2026), written right here in Berlin with Marc Toussaint (TU Berlin) and Wojciech Samek (Fraunhofer HHI). A rare chance to ask the author directly why things were done the way they were. In simulation a robot can cheat: it knows exactly where every object is. Policies trained with this privileged state learn contact-rich behaviour well, but a real robot only gets noisy camera images. The usual fix is teacher-student distillation, where a vision policy learns to copy the expert’s actions. That forces the student to do two jobs at once: work out the state from pixels, and relearn a mapping from state to action that the expert already has. This paper keeps the expert and only learns a visual state estimator. The catch is that an error in the estimate can matter a lot or not at all, depending on what the expert does with it. So the estimator is trained on two signals: plain state regression, and an action-consistency loss backpropagated through the frozen, differentiable expert. Across five goal-conditioned manipulation tasks, keeping the expert consistently beats direct pixel-to-action imitation trained on the same demonstrations. On a real Panda arm it reaches 76% success, without retraining the expert. No preparation needed. Come with questions for the author, disagreements or your own view on end-to-end versus modular. 🎟️ There are only a limited number of spots, so we can keep the session personal and interactive. Please register early and only attend if you’ve received a ticket. 📍 Hosted at the Merantix AI Campus on November 23rd. Doors open 18:30, we start at 18:45. 👋 New here? Join the community group - everything else gets announced there first.

13 going via Luma
Add to calendar
⋯
THU 26Nov
Thu 26 Nov 2026, 18:00 Merantix AI Campus Merantix AI Campus Berlin

Building Today for Tomorrow’s Impact on Healthcare When Clinicians Build: Showcase and Open Floor Healthcare changes when the people working in it decide to fix what they run into every day, and build solutions that work far beyond their own ward. This edition of the Digital Health Cocktail Hour puts those people on stage. Physician-led and researcher-led teams from the BIH Digital Health Accelerator at Charité will show what they’re building and, just as importantly, what is still standing in their way. Each team ends with the questions they’re wrestling with. Then it’s over to the room: founders, investors, clinicians and operators, bringing their own experience to the conversation. Program Project showcase: Translational projects from the BIH Digital Health Accelerator at Charité, led by treating physicians and researchers, have five minutes each to present their work. Every team closes with the questions they most want to discuss – whether their data can answer the clinical question, for instance, or what it takes to move from a research tool to a product. Projects include: → CALA-XR: The limb is gone, but the pain is real. Therapy for phantom limb pain, built for one patient at a time, in extended reality. → OnKIboard: A cancer diagnosis can’t wait. Neither can the treatment decision. An AI agent system that aims to give tumour boards fast, valid and traceable treatment decisions. → CellTracer: Bringing a cell therapy to patients takes more than science. A regulatory-intelligence platform for cell therapy that helps teams identify comparable trials and regulatory precedents. → CoRehApp: Intensive care ends at discharge. The road back from its after-effects doesn’t. A personal avatar companion that supports recovery and connects patients with the care they need. → FlowPilot: Every blood cancer case starts as a raw file and ends as a report. Faster, more consistent interpretation, informed by similar past cases, with clinicians keeping the final call.

via Luma
Add to calendar
⋯
MON 30Nov
Mon 30 Nov 2026, 18:30 Merantix AI Campus BRISA Berlin

This Monday we are looking at VLAW: Iterative Co-Improvement of Vision-Language-Action Policy and World Model (Guo, Lee, Shi, Chen, Liang & Finn, 2026) from Stanford and Tsinghua, presented at ICML 2026. Two weeks ago, Long-WAM built prediction into the policy itself. VLAW keeps policy and world model separate and asks whether each can make the other better. Improving a robot policy through practice means running it again and again on real hardware, which is slow and expensive. A learned world model could stand in as the simulator. However, today’s world models are trained mostly on successful demonstrations: they are too optimistic and they miss the small physical details that decide whether a contact-rich task works. VLAW closes the loop, by rolling out the policy (π0.5) on a real Franka arm (failures included). Those rollouts finetune the world model. The policy then practises inside the improved world model at scale, a vision-language model judges which imagined attempts succeeded and the policy is finetuned on the successful real and synthetic ones. Across contact-rich tasks like scooping, erasing marks and stacking, this improves success by 39.2 percentage points over the base policy, with 11.6 of those coming from the synthetic rollouts. As always: No preparation needed. Come with questions, disagreements, or your own take on how much a robot can learn from its own imagination. 🎟️ There are only a limited number of spots, so we can keep the session personal and interactive. Please register early and only attend if you’ve received a ticket. 📍 Hosted at the Merantix AI Campus on November 30th. Doors open 18:30, we start at 18:45. 👋 New here? Join the community group - everything else gets announced there first.

via Luma
Add to calendar
⋯
23 NovTomorrow
Register →