Zurich AI, ML and Computer Vision Meetup - July 30, 2026

Join our in-person meetup to hear talks from experts on cutting-edge topics across AI, ML, and computer vision.

Data curation is often described as drudgery, but an agent can turn it into an interactive, puzzle-solving process. This talk walks through a data-centric computer vision workflow built on the FiftyOne platform and the FiftyOne Agent, which uses skills (step-by-step instructions for tasks) and the Model Context Protocol (MCP) to handle dataset management, model inference, and brain computations through natural language — querying metadata, visualizing class distributions, and running complex operations without writing manual code.

The demo applies the workflow to a crop and weed dataset of 8,034 images and more than 112,000 labels (bounding boxes, keypoints, and segmentation masks). The agent diagnoses severe class imbalance and duplicate frames, builds interactive histograms and dashboards to surface domain shifts like lighting variation, finds invalid annotations so they can be corrected directly in the FiftyOne App, then automates exporting the curated dataset, fine-tuning a YOLO model, and evaluating its performance. The result is a repeatable pattern — ingest, explore, improve, fine-tune, evaluate, iterate — with a human in the loop through an intuitive, agentic interface.

Independent authors struggle to reach younger audiences in the short-form video era, yet typical AI video generation remains a "prompt and pray" manual process. This session introduces ReelaTales, an agentic multimedia production pipeline that acts as an autonomous AI director to convert static manuscripts into cinematic, social-ready video trailers.

We will dive into the architecture of our orchestration fleet, detailing how we moved beyond third-party dependencies to leverage Apertus, the open-weights Swiss LLM, running on local infrastructure to handle semantic text analysis, emotional beat mapping, and cinematic shot choreography. Attendees will walk away with practical insights on chaining local foundation models with heavy video APIs to maintain strict narrative integrity and character consistency across automated production pipelines.

Automated visual inspection represents a massive market opportunity with the profound potential to save lives. In this talk, I will highlight recent success stories driven by algorithmic advancements in foundation models for visual inspection. I will then explore two real-world use cases (medical diagnostics and civil infrastructure monitoring) that IBM Research has tackled over the past few years.

Finally, I will conclude by outlining the remaining open challenges in deploying these systems to inspire future work within the community.

Traditional Agentic RAG is great when we know what we’re looking for, or when the LLM knows how to plan the task. But what happens when our questions sound more like: Does a falling apple have anything to do with why we walk? Would AI, specifically Agentic RAG, know where to begin?

LLMs are promising, but they still have limits: they can over-rely on semantic similarity and miss the strange, indirect connections that often drive discovery. Innovation means navigating new territory, where the unknown is not a bug but the constant.

How do we use the technology we have to explore what we don’t yet fully know? How do we build AI systems that help us get deep insights into the topic we are interested on.