Meetup topic
Artificial Intelligence
Date
7 May 2025
Location
El Niño office in Enschede
Speakers
Mauk Muller (El Niño)
Terhi Lehtola (Vaisala)
Anand Chowdhary (Sycamore)
The AI Meetup
In May, El Niño proudly hosted the AI Meetup, an evening dedicated to data-driven AI projects and the trends shaping the industry. The event brought together data scientists, AI specialists, engineers, and enthusiasts who gathered to learn, connect, and dig into the technical realities of building AI today.
Why We Did It
We organized this meetup because AI has moved from hype to practice. Everyone is talking about what AI could do, but the more interesting conversations are about what teams are actually shipping: the sensors collecting the data, the systems running models in production, and the hard questions of how you measure whether a model is any good.
At El Niño, we work on these problems every day. With the A a space for three in-depth technical talks, notsurface-level overviews, so practitioners could share what they learned along the way, followed by networking andrefreshments.Inspiring TalksThe evening’s lineup showed AI in practice, from the sensork:
Terhi Lehtola (Vaisala) spoke about smart sensors, the foundation of any data-driven system. She showed how quality measurement at the edge shapes everything downstream, and why good AI starts long before the model with reliable, well-understood data.
Anand Chowdhary (Sycamore) shared lessons from running AI systems in production. He went beyond the demo stage to the real engineering challenges of reliability, scale, and keeping a system trustworthy once real users depend on it.
Mauk Muller (El Niño) talked about LLM benchmarking, tackling the question everyone in the room cared about: how do you actually know if a model is performing? He walked through how El Niño evaluates large language models and why honest measurement matters more than leaderboard scores.
Takeaway
The three talks formed a clear arc, from raw data to deployed model to honest evaluation. Good AI starts with good data: before any model, there's a sensor, a measurement, a pipeline, and reliability at the source decides the ceiling for everything downstream. Production is where the hard problems live; a working demo is the easy part, while scale, reliability, and trust are the real engineering work once users depend on the system. And honest measurement matters more than a high score, because thoughtful benchmarking is how you separate real performance from hype.
The bigger picture is that AI is no longer about whether it can be done, but about doing it well, responsibly, and at scale. The conversations that carried into the networking afterwards proved how much appetite there is in the region for exactly this kind of deep, practical exchange.
Thanks to everyone who joined, and to our speakers for sharing the real work behind the headlines. We'll be back with more!
"To learn anything, you have to become comfortable with getting outside of your comfort zone."
- Product Owner and Operations Lead at El Niño
