Digital Crop Scientist · ML Engineer
Nelson Pinheiro
Crop scientist and ML engineer based in Bonn, currently finishing my M.Sc. at the University of Bonn while working as a Research Assistant at the Agrobotics Group and supporting Sales & Marketing at DynamoBot GmbH. From October I start a PhD at the Institute of Geodesy and Geoinformation (IGG) at the University of Bonn.
Pinheiro is Portuguese for pine tree. I’ve spent my career studying everything that grows.
Who I Am
In My Own Words
Crop scientist and ML engineer based in Bonn, currently finishing my M.Sc. at the University of Bonn while working as a Research Assistant at the Agrobotics Group and supporting Sales & Marketing at DynamoBot GmbH. From October I start a PhD at the Institute of Geodesy and Geoinformation (IGG) at the University of Bonn.
My thesis focuses on precision plant phenotyping from UAV-acquired RGB orthomosaic imagery, using YOLO-based object detection to locate individual maize plants, measure row spacing and inter-plant distances, and produce RTK-accurate georeferenced plant maps. Before the M.Sc., I completed a B.Sc. in Agricultural Science with a plant science focus, studying cation transporters in abiotic stress tolerance.
Before university, I completed a full vocational training as a Bankkaufmann (Bank Clerk) at KfW Bankengruppe, rotating through Corporate Finance, Environmental Finance, Credit Risk, and Education Finance. It’s an unusual background for a crop scientist, but it means I can read a balance sheet, assess project risk, and communicate with the business side of AgTech, not just the lab side.
My work combines deep domain knowledge in plant and soil science with modern software engineering. I build ML pipelines, geospatial platforms, and IoT sensor systems that solve real problems in agriculture.
Timeline
Experience
Sales & Marketing
ActiveAgTech startup developing autonomous weeding robots (WeWeed, WeSee, WeMan). Supporting go-to-market strategy and customer engagement, bridging technical capabilities with market needs.
Research Assistant (Wissenschaftliche Hilfskraft)
ActiveOperating robotic phenotyping platforms for agricultural research. Deploy and run the TIPaRD field platform at Campus Klein-Altendorf, acquiring geo-referenced, time-series RGB and multispectral crop imagery for growth modelling across field plots, plus a rail-based greenhouse system for red bell pepper and tomato phenotyping with structured image datasets for trait analysis. Built and optimised ML models predicting soil water content and nitrogen from Vis-NIR spectral data, run drone-based aerial capture as a licensed drone pilot, and represent the group at industry fairs such as Agritechnica.
Administrative Officer
Education finance information center. Part of a 3-year vocational training as Bankkaufmann rotating through Corporate Finance, Environmental Finance, Credit Risk Management, and Education Finance, building a financial and analytical foundation that now informs how I evaluate and communicate AgTech projects.
Timeline
Education
PhD (coming soon)
Doctoral research in remote sensing and geoinformation. More detail once the research focus is settled.
M.Sc. Crop Science (Digital Focus)
ActiveThesis: Maize plant detection from UAV RGB orthomosaics using YOLO, comparing annotation strategies (bbox, OBB, point) to derive row spacing, inter-plant distance, and RTK-accurate individual plant maps.
B.Sc. Agricultural Science (Plant Science Focus)
Thesis: Characterisation of the role of cation transporters in plant abiotic stress tolerance. Strong foundation in plant physiology, experimental design, and statistical analysis.
Vocational Training: Bankkaufmann (Bank Clerk)
Full 3-year accredited vocational training. Rotations: Corporate Finance · Environmental Finance · Credit Risk Management · Education Finance. A deliberate detour that built financial literacy, stakeholder communication, and project assessment skills.
Principles
How I Work
Scientific First
I start with the science. Understanding the biological system is prerequisite to modeling it effectively.
Iterative Development
I build incrementally, validating at each step. Complex systems are understood through simple models first.
Reproducibility by Default
Every analysis is version-controlled and documented. If I can’t reproduce it, it’s not done.
Full-Stack Thinking
From embedded sensors to cloud dashboards, I build complete systems that work end-to-end.
Have a Project in Mind?
I take on a small number of consulting projects alongside my PhD research. Crop imagery pipelines, phenotyping platforms, and spectral data modelling are where I add the most.