Projective Duality

Jan Saputra Müller

I'm a long-time researcher, data scientist, consultant and entrepreneur in the field of Machine Learning/AI. My professional journey in that field started in 2008 when I became a researcher at TU Berlin's ML group. Nevertheless, I have been passionate about math, computer science and philosophy since I was a young kid and it never changed. My journey in those fields started when I was 8 years old when I started to learn coding in C. I experimented with self-implemented neural networks already in high school.

During my time as a full-time researcher at the Machine Learning Group at TU Berlin, my research focused mainly on time series analysis, in particular blind source separation (Stationary Subspace Analysis, thesis) and co-adaptive learning. For the latter, I developed the first purely mathematical formulation.

Since 2012, I'm part of Berlin's entrepreneurial scene. As idalab's first consultant, I solved many practical problems for customers using data science and mathematical modeling. Customers included very early stage startups, but also large companies from the financial and pharmaceutical industries. Some of the projects that started back then developed massively and are still going on today.

Since 2014, I co-founded several startups, all of them with Machine Learning at their core. One of the products I'm the most proud of was a natural-language-to-database-queries technology that worked astonishingly well — and that was about 6 years before ChatGPT. The most important companies that I co-founded are:

• SevenPlans (2014): The first tool that was able to measure the real impact of TV and out-of-home advertising. We developed a Bayesian attribution model that we called the "echo model", which is based on the idea that advertising spendings translate into characteristic "echos" in the sales timeseries.
• PubStation (2016): A platform designed to help app developers monetize apps on connected TVs. It was an early bet on the Smart TV app economy and provided valuable product and market insights during a phase when the ecosystem was still maturing compared with smartphones.
• AskBy (2017): A software that allows people to ask questions to a database in plain English. Using state-of-the-art natural language processing and an own developed sequence-to-sequence model called Nefisto (neural finite state output), AskBy was able to perform such tasks with a high accuracy almost 6 years before ChatGPT.
• MorphAIs (2019): Originally founded as a data-driven venture capital firm to make better investment decisions by reducing human bias and increasing rational decision-making. Later, the company evolved into a service provider and was acquired by Evertrace in 2025.

I'm also passionate about Bitcoin. Already interested in it since 2012, I'm still learning almost every day about this very complex and exciting topic. At this point, I'm already experienced in developing smart contracts with Bitcoin Script and some other smart contract languages. I would say I understand the technology on a semi-pro level now — not as much as some people from the Bitcoin dev team, but very close to it. Of course, apart from the technology, I'm very interested in the social and economic/monetary impacts of Bitcoin.

Outside of work I love being in nature and connecting with new people. The universe, life and consciousness are mysteries that I'm very fascinated by. I love to philosophize and going down that rabbit hole. Let's talk about how we can create a better future for this world! :)