Across Germany, developers working hands-on with generative AI noticed the same thing: there aren’t many spaces for deep, technical conversations about LLM applications. That’s why the AIDev community was created — to connect those building with LLMs, solving hard problems, and pushing the field forward.

After three successful events, AIDev is coming to Berlin for the first time – hosted at ⭕️KI Park, in cooperation with the KI Bundesverband, AI Village, Deyan7 GmbH & Co KG, and ellamind.

🧠 What to expect:

🔹 LLM Inference Optimization – Dennis D. (Seedbox)will do a technical deep dive and explain how sparse pruning can increase the performance of the model at inference time and how their tailormade prompt router distributes prompts to different models and finetunes based on the task.

🔹 Traceability in Multistep LLM Systems – Marc Klingen (Langfuse (YC W23)) will give us a deep dive into advanced techniques and best practices for implementing traceability and observability in multi-step LLM systems.

🔹 Scaling LLMs: Intelligent Routing & Performance Tracing – Lucas Hild (Knowunity)and his team have built one of the fastest-growing AI education platforms, now serving over 18 million users. He’ll share insights into their custom model-routing strategies, their approach to evaluating LLM performance, and key learnings from deploying large-scale language models at scale.

🔹 Operating the largest TaxAI Plattform in Germany – Learnings from the Journey – tba

🔹 Bring your own problem – Share your current challenge during registration and we’ll tackle it live with the speakers.

📍 15 May | 17:30–22:00 | at KI Park Berlin

This event is for developers and researchers working with LLMs and generative AI on a daily basis. To attend, please share your GitHub or Hugging Face profile and a short description of a problem you’re currently solving.