Medical AI | Validated Deep Learning | Vienna

Jacob Pichelmann

Head of AI Development at RetInSight

I build medical AI that holds up in the clinic.

At RetInSight, a Medical University of Vienna spin-off and part of Topcon Healthcare since 2025, I lead the AI engineering team that takes models for retinal imaging from data and clinical annotation to validated, certified, deployed product.

I came to medical AI through econometrics. I trained at the Toulouse School of Economics, graduating top of my track, and did empirical research at the OECD and the University of Zurich. That grounding in measuring things rigorously still shapes how I build and evaluate models.

Experience & education

A path that converges.

Economics, research, law and building things on the side, all merging into what I do now. Select any entry for the details.

in progress

2026 – now
2025 – 2026
2022 – 2025
2020 – 2021
2019 – 2020
2015 – 2018
2013 – 2014
2011

Work | Jun 2026 – present

Head of AI Development

RetInSight | Vienna, Austria

Medical University of Vienna spin-off building AI for retinal OCT analysis. Part of Topcon Healthcare since May 2025.

Since June 2026 I lead the AI engineering department, with five direct reports and full ownership of how we develop AI.

My job is to keep our development engine running: a team that consistently delivers state-of-the-art models and methods in retinal imaging, a framework that scales with what we will build next, and a clear strategic direction for our AI work.

That includes shaping our regulated development process under MDR and the EU AI Act, representing AI development in audits, and working closely with clinical, regulatory and product colleagues.

Selected work

AI screening for retinal disease: from algorithm to clinical evidence

retinsight.com

Side projects | Aug 2026 – present

Up! More time with friends

Side project

A personal project built around a small idea: make it easier to actually meet friends by letting them say on which day they are up for something.

I built it with today’s agentic coding tools and shipped it to the App Store and Google Play, where around 200 people now use it.

Selected work

Up! More time with friends

Work | Feb 2025 – Jun 2026

AI Development Lead

RetInSight | Vienna, Austria

Medical University of Vienna spin-off building AI for retinal OCT analysis. Part of Topcon Healthcare since May 2025.

Alongside my work as a developer, I took on responsibility for how we work: refining parts of our machine-learning lifecycle process and leading selected projects both organisationally and technically, from planning through to delivery.

In this time I led the development of an AI screening algorithm for retinal biomarkers on OCT, and the reader study showing that it helps clinicians make better decisions, which I presented at ARVO 2025 and 2026.

First author at ARVO 2025 & 2026
Acquired by Topcon, May 2025

Selected work

AI screening for retinal disease: from algorithm to clinical evidence

ARVO abstracts

Improving community-based OCT screening through real-time AI-based biomarker visualization ARVO 2026 | IOVS 67(7):331 | first author
AI-based OCT screening for retinal biomarkers: results from large real-world data ARVO 2025 | IOVS 66(8):5462 | first author
Explainability vs. data efficiency: comparing specialized and foundation models for retinal feature classification ARVO 2026 | IOVS 67(7):5661
Evaluating the robustness of an OCT biomarker detection model in a predominantly healthy cohort ARVO 2026 | IOVS 67(7):5660
Prevalence and characterization of pigment epithelial detachment and hyporeflective spaces detected on OCT using AI in optometry ARVO 2026 | IOVS 67(7):922
A multi-stage filtering framework for large-scale cohorting of patients with diabetic macular edema ARVO Imaging in the Eye 2026 | IOVS 67(9):PB0050

retinsight.com

Work | Feb 2022 – Feb 2025

Deep Learning Engineer

RetInSight | Vienna, Austria

Medical University of Vienna spin-off building AI for retinal OCT analysis. Part of Topcon Healthcare since May 2025.

I joined RetInSight as a junior, supporting model development and improving how we evaluate our models. Within a year I took ownership and technical leadership of one algorithm, biomarker segmentation on scans from Triton OCT devices, and later of segmentation models in further domains.

With the support of colleagues, I was responsible for the whole lifecycle: data sourcing and our in-house annotation process, model optimisation, validation (which matters enormously in a clinical setting, from writing the validation plan to running it), and packaging models so they run reliably in production rather than only in an experiment.

Promoted after 10 months
First author at ARVO 2024

Roles

Deep Learning Engineer Dec 2022 – Feb 2025
Junior Deep Learning Engineer Feb 2022 – Dec 2022

ARVO abstracts

Navigating the complex landscape: identification of clinical and subclinical features that affect the reliability of fluid biomarker measurement ARVO 2024 | IOVS 65(7):3759 | first author
Can AI provide more consistent fluid measurements in nAMD across OCT devices than humans? ARVO 2024 | IOVS 65(7):5636

retinsight.com

Work | Nov 2020 – Aug 2021

Research Assistant

University of Zurich | Remote

Department of Economics, Kühne Center.

Alongside my master’s, I worked remotely as a research assistant at the University of Zurich’s Kühne Center with Dr. Mathilde Le Moigne, building a dataset of around 40 million trade and macroeconomic observations.

The analysis was presented in the Kühne Center Impact Series and became the starting point for my master’s thesis with Prof. Ralph Ossa.

Selected work

Moving beyond efficiency: a novel framework to measure trade-flow resilience

Education | Sep 2019 – Sep 2021

MSc Econometrics & Empirical Economics

Toulouse School of Economics | Toulouse, France

Ranked #10 economics department worldwide and #3 in Europe (RePEc, 2026). Home of Nobel laureate Jean Tirole.

I chose Toulouse because it is one of the best places in continental Europe for quantitative economics, and it lived up to that: two extremely challenging but rewarding years, in which I learned as much from my brilliant classmates as from the faculty.

I specialised in econometrics, focusing on statistical modelling, empirical analysis, high-dimensional models and forecasting, and in my second year I finished top of my track as valedictorian.

Valedictorian, Econometrics track
Highest honours
TSE scholarship 2020

Selected work

Moving beyond efficiency: a novel framework to measure trade-flow resilience

tse-fr.eu

Side projects | Oct 2021

Sustainability Data Science Challenge

Co-hosted by Capgemini and Microsoft

A challenge co-hosted by Capgemini and Microsoft: estimate CO₂ emissions, biodiversity and solar potential from satellite images. I took part together with a colleague from university.

We built a Keras segmentation pipeline and designed custom augmentation to rebalance the skewed training data, which improved performance by 80% and took us to first place for model performance and second overall.

1st for model performance
2nd overall

Selected work

Sustainability Data Science Challenge

Work | Mar 2019 – Mar 2020

Intern, then Consultant

OECD | Paris, France

Economics Department.

At the OECD Economics Department in Paris I worked on a question that affects millions of households: why people in some countries move home far more readily than in others.

I did the full statistical analysis and methodology on household microdata from Europe, the US and Australia, co-authored the resulting working paper, and presented the findings to policy stakeholders and OECD country desks.

Co-author, OECD Working Paper No. 1626

Selected work

Should I stay or should I go? Housing and residential mobility across OECD countries

Work | May 2020 – Jul 2020

Data Science Intern

Erste Group | Vienna, Austria

During my master’s I spent a summer at Erste Group, helping build a production machine-learning pipeline for detecting financial crime.

It was my first close look at what running ML in production really takes: I audited the data quality and defined the indicators used to monitor it.

Education | Oct 2015 – Nov 2018

BSc Economics

WU Vienna | Vienna, Austria

Vienna University of Economics and Business.

I started out studying technical physics at TU Wien. I loved the maths, but its application felt too abstract for me, and economics offered exactly the combination I was looking for: a statistical and mathematical foundation, applied to real questions.

At WU I graduated with highest honours in the top 1% of my cohort, and my bachelor’s thesis, testing the efficient market hypothesis with neural networks, was my first deep learning project.

Highest honours
Ranked 30th, top 1%
Dean’s List
Merit scholarships 2016–2018

Education | Oct 2015 – In progress

Law (Mag. iur.)

University of Vienna | Vienna, Austria

I have been studying law at the University of Vienna alongside economics and, later, my career, and I am still working towards the degree.

The legal training pays off in my day job, where reading and implementing medical-device regulation and the EU AI Act is part of building clinical AI.

Side projects | Mar 2015 – Jun 2018

Co-founder

Findus | Vienna, Austria

At the start of my studies, two friends and I, a very talented programmer and a very talented UI/UX designer, saw something missing in how people discover events and decided to build it ourselves.

With support from the Austrian Economic Chamber (WKO) we launched and saw promising early user growth, but personal reasons and competition led us to wind it down. We had what it took to build the product, not yet the business experience to navigate the market, and it was an incredibly rewarding journey that taught me a lot.

Business plan, 2017 (PDF, German)

Work | Oct 2013 – Jun 2014

Civilian service

Verein Ute Bock | Vienna, Austria

A Viennese organisation supporting refugees and asylum seekers.

I did my civilian service, Austria’s alternative to military service, at Verein Ute Bock, which supports refugees in Vienna. It was the year before the refugee crisis reached Europe, and the work was as demanding as it was meaningful.

I was responsible for event management and fundraising, organising a series of benefit events that raised a record €45,000 in donations, and I helped refugees with legal questions, backed by experts, which meant regular contact with the authorities.

Record €45,000 in donations

Work | 2011

Editorial intern

DIE ZEIT

One of Germany’s most renowned weekly newspapers.

While still at school, I did an internship at DIE ZEIT, one of Germany’s most renowned weekly newspapers.

Together with Florian Gasser I wrote “Fürs Leben gezeichnet” for the Austria section, a feature on prison tattoos as a forgotten piece of folk culture, published in September 2011 with my name in the byline.

Article: “Fürs Leben gezeichnet” (PDF, German)

Selected work

Things I’m proud of.

Clinical evidence, economic research and a few things I’ve built on the side.

Jacob Pichelmann

A path that converges.

Things I’m proud of.

AI screening for retinal disease: from algorithm to clinical evidence

Should I stay or should I go? Housing and residential mobility across OECD countries

Moving beyond efficiency: a novel framework to measure trade-flow resilience

Sustainability Data Science Challenge

Up! More time with friends