applydata by diconium · 2024 – Present
applydata by diconium · 2024 – Present
applydata is the data and AI community arm of diconium — a space where practitioners, engineers, and product thinkers come together to share what they're building, what they're learning, and where the industry is heading. The community expresses itself through a blog covering data engineering, AI, and automotive tech, a growing events programme spanning multiple cities, and a product suite that includes AI assistants, segmentation tools, and smart search solutions.
The tagline captures it well: "We build what's next. We share what we learn." My contribution sits at the intersection of both — helping shape the content that reflects the community's expertise, and creating the in-person moments where that expertise becomes a conversation.
One of the ways I contribute to applydata is through editorial content — writing articles that translate complex technical subjects into accessible, thought-provoking reads for a data-savvy audience. This requires the same kind of structured thinking I apply in product work: understanding who the reader is, what they already know, and what insight they should walk away with.
Published Article · January 2026
Is Gamified Augmented Reality the Future of Automotive UX?
Exploring how AR head-up displays, gamification mechanics, and AI-powered interfaces are reshaping the in-cabin driving experience — and what that means for automotive UX design going forward.
Published Article · September 2024
Data Labeling as a Continuous Service
Making the case that data labeling is not a one-time project but an ongoing, evolving service — tracing the history of ADAS and autonomous driving to explain why continuous, high-quality labeled data is the foundation modern ML models are built on.
Published Article · March 2026
Sycophancy in AI Models: When Your AI System Is Optimized to Agree With You
Why AI systems trained on human feedback are systematically biased toward agreement — and why that's a business risk hiding inside your approval metrics. Grounded in three 2025 research papers, this piece bridges academic findings on RLHF and the practical consequences for companies deploying AI at scale.
Each article draws directly on my day-to-day work — spanning automotive UX, AI tools for practitioners, and the evolving role of data in ADAS development. The topics are deliberately varied, but the approach is consistent: find the angle that adds something to the conversation, not just summarise what already exists.
applydata runs a rich events programme — from focused technical meetups on data engineering, AI, and cybersecurity, to the annual applydata summit and the hands-on applydata garage AI barcamp format, where practitioners team up to tackle real AI problems in a collaborative session. The events are a core expression of the community's values: expert-led, practically focused, and genuinely useful to the people who attend.
My role in this has been to help grow and sustain that programme — supporting campaign planning, coordinating communications, and making sure each event has the content and context it needs to land well with its audience.
🇷🇴 Milestone — First applydata Meetup in Romania
In 2025, I organised the first applydata meetup ever held outside Germany — bringing the community to Bucharest for an event focused on cybersecurity in automotive, including a live AI audit demo. This was a meaningful expansion of the applydata footprint: taking a community that had been rooted in the German tech scene and planting it in a new market, with a local audience, local speakers, and local energy. Planning and executing this required coordinating across teams in multiple countries, managing logistics from scratch in a new venue city, and building the local audience from the ground up.
Community and content work might seem like a departure from product management — but in practice, it exercises many of the same muscles. Defining what story to tell, for whom, and in what format is a product decision. So is deciding which events to prioritise, how to grow an audience incrementally, and how to measure whether what you're putting into the world is actually creating value for the people on the receiving end.
This work also keeps me close to the broader data and AI conversation — the trends, the debates, the tooling — which directly informs how I think about product problems in my ADAS work and beyond.