Market Path — People behind the seminars

The team that runs these seminars

Four specialists with backgrounds in machine learning research, instructional design, and applied neural network engineering. Each brings a specific perspective to how the material is taught and discussed.

Market Path seminar environment showing collaborative neural network study session

Who runs the programme

Each person on this team holds a defined role. There is no overlap by design — the curriculum, research facilitation, and participant experience are handled separately.

Dmitri Falck, Lead Instructor for neural architecture seminars at Market Path

Dmitri Falck

Lead Instructor — Neural Architecture

Dmitri spent eight years working on transformer-based models before moving into education. He structures each seminar around concrete architectural decisions rather than theoretical overviews. His sessions tend to involve a lot of whiteboard-style reasoning through real model failures.

Soraya Nkemdirim, Curriculum Designer at Market Path

Soraya Nkemdirim

Curriculum Designer

Soraya sequences the material across each seminar series. She decides what participants encounter first, how concepts are layered, and where discussion breaks are placed. Her background is in cognitive load research, which shapes how densely each session is packed.

Tobias Vrabel

Research Facilitator

Tobias selects the papers and case studies that feed into each seminar. He reads across convolutional architectures, attention mechanisms, and training dynamics to find material that is recent enough to be relevant but stable enough to teach. He also moderates the deeper technical discussion threads between sessions.

Ines Quartermaster

Participant Experience Lead

Ines handles everything that happens outside the actual sessions — onboarding, scheduling across time zones, access to materials, and follow-up after each cohort ends. She also collects structured feedback and passes it directly to Soraya for the next curriculum revision.

What the team has delivered

Numbers from the programme since it launched in 2017. These reflect completed cohorts and active participants across all seminar tracks.

38 Completed seminar cohorts
12 Countries represented by participants
6 Distinct architecture tracks taught
Transformer architecture Convolutional networks Attention mechanisms Training dynamics Recurrent models Graph neural networks Model interpretability Distributed training