NextPrevious

Session

C | Scalable Data Stream | Solution Study

Monday, March 17

11:45 AM - 12:15 PM

Live in Berlin

Less Details

Data sourcing for ADAS/AD training and validation remains a pain: Cost, delivery times and quality present bottlenecks for training and validation. Automation is an exciting promise, but how can it be implemented efficiently and reliably? Kognic presents results from a recent productivity-enhancement effort, opening up for discussion.

In this session, you will deep dive into keytopics such as:

  • Accelerating speed with auto-labels in a counter-intuitive way
  • Using the right tools, the right way: designing for desired supplier behavior
  • Annotator goal setting and up-skilling: what does and doesn’t work
  • Systematic quality gate design for reliable automation in QC
  • From problems to better problems: resulting challenges for data sourcing
Presentation

Speaker

Olof Fyr

Product Manager, Kognic

Olof Fyr is a Product Manager at Kognic, specializing in annotation efficiency and automation for ADAS and autonomous driving. With a Master’s in Physics focusing on Complex Adaptive Systems, he started as a machine learning engineer working on problems like speech recognition, computer vision, and SLAM. For the past five years, he has focused on optimizing data annotation, bridging automation and human performance to improve AI training and validation.

The Pop in Your Job – What drives you? Why do you love your job?
What excites me most about my work is shaping the future of AI. The combination of technical depth (machine learning, automation, and data pipelines) and the human side (psychology, training, and behavior design) creates a uniquely complex and rewarding problem space. I love tackling challenges that blend these dimensions, driving innovations that make AI development more efficient and scalable. By improving how data is sourced and labeled, I get to contribute directly to the advancement of autonomous systems.

Company

Kognic

Kognic provides a modular MLOps platform to ADAS/AD perception developers. It includes sensor fusion annotation and analytics tools, enabling automotive-grade development of performance-critical ML models.

NextPrevious