About

Welcome to my blog. I write about coding, data projects, and sometimes about my personal life.

My professional story

I am interested in building exciting technology with great teams.

I have spent some years building products that use deep learning to understand human motion in video: Anything from video-powered workout apps to gesture recognition for the automotive industry.

People often forget that building these AI products requires a ton of “boring old software” on the side: Web apps for data management, build pipelines for deliverables, deployment of web services. This has now led me back to a broader space of simply building good software – with or without AI inside.

Working in both product and service companies over the years has taught me to put the user at the center of every product decision – a good reminder when working with technology.

---

Vita

When	Position	Location
Since 2024	Engineering Lead at they Consulting	Hamburg
2021–2024	AI Lead at LifeBonus	Hamburg
2017–2021	AI Engineer (and later Team Lead) at Twenty Billion Neurons	Berlin
2017	M.Sc. in Computer Science from University of Hamburg	Hamburg
2016	Co-founded Album Up, a web service for photographers	Hamburg
2013–2017	Web developer (part-time) at YOOtheme	Hamburg
2012–2013	Web developer at Showcase Software Ltd	Wellington, NZ
2012	B.Sc. in Computer Science from University of Hamburg	Hamburg

Publications

I have authored and been co-author on a few publications about Machine Learning and Computer Vision (most recent listed first)

What to Say and When to Say it: Live Fitness Coaching as a Testbed for Situated Interaction (2024)

Citation: Panchal, S., Bhattacharyya, A., Berger, G., Mercier, A., Bohm, C., Dietrichkeit, F., … & Memisevic, R. (2024). Live Fitness Coaching as a Testbed for Situated Interaction.

• Link to paper

Is end-to-end learning enough for fitness activity recognition? (2023)

Citation: Antoine Mercier, Guillaume Berger, Sunny Panchal, Florian Dietrichkeit, Cornelius Böhm, Ingo Bax, and Roland Memisevic. Is end-to-end learning enough for fitness activity recognition? CoRR, abs/2305.08191, 2023.

• Link to paper

Localizing salient body motion in multi-person scenes using convolutional neural networks (2019)

Note: Still published under my birth name “Florian Letsch”

Citation: Letsch, F., Jirak, D., & Wermter, S. (2019). Localizing salient body motion in multi-person scenes using convolutional neural networks. Neurocomputing, 330, 449-464.

• Link to paper

---

Selected projects