From research to practice

Systems & Projects

Connecting models, data, and physical systems to answer useful questions.

Scientific imaging · AI integration

Entomoscope & ENIMAS

From specimens to usable images: I lead AI development and integrate analysis into the Entomoscope imaging workflow.

Problem: Digitizing small insects requires controlled imaging and a practical connection between hardware and analysis.
My contribution: I lead AI development for the Entomoscope photomicroscope and contribute to its integration with the DiversityScanner robotic ecosystem.
System / method: ENIMAS connects image acquisition with analysis workflows for species identification, orientation, sex classification, and anatomical segmentation.
Evidence: The Entomoscope 2.0 / ENIMAS 2.0 preprint describes the platform. Related specimen-triage research evaluates downstream analysis tasks.

Computer vision · Biological measurement

Measurements from images: I develop software to estimate insect length, volume, and biomass for ecological research.

Problem: Manual measurements limit the scale of insect trait and biomass studies.
My contribution: I develop the image-based software suite and deep learning workflows for insect morphometric analysis.
System / method: Oriented bounding boxes support rapid measurement; anatomical segmentation supports more detailed, taxon-specific estimates from 2D images.
Evidence: Published in Ecological Informatics in 2026, with taxon-specific volumetric validation. Estimates depend on the imaging setup and the taxa evaluated.

Fine-grained recognition · Forest health

Vision for forest surveillance: I develop classifiers for regulated forest pests within the FORSAID project.

Problem: Recognizing regulated pests in trap images is difficult when species look similar and imaging conditions vary.
My contribution: I contribute deep learning classifiers to work package 3: digital technologies for ground detection and surveillance of regulated pests.
System / method: Image-based species recognition connects computer vision with expert taxonomic knowledge in the Horizon Europe FORSAID project (grant 101134200).
Evidence: The Agrilus recognition study investigates neural networks for forest pest surveillance, including evaluation on specimen images.

Robotics · Biodiversity workflows

Connecting specimen handling, imaging, and AI-assisted analysis.

Problem: Small, fragile biological specimens are difficult to handle and document consistently.
My contribution: I contribute AI analysis and collaborate with engineers and taxonomists to connect robotic imaging with specimen-level information.
System / method: The broader research spans flexible gripping, close-range imaging, orientation and sex classification, and anatomical segmentation.
Evidence: Related publications cover flexible gripping, specimen triage, and acoustic levitation for multi-view imaging.

Industry · Decision support

Optimization and data tools for industrial maintenance workflows.

Problem: Industrial maintenance planning requires usable decision support and consistent data workflows.
My contribution: As a Data Scientist Consultant at Modis / Baker Hughes, I developed a turbine-maintenance optimization algorithm, led cloud data migration, and built workflow automation tools.
System / method: Optimization, data migration, and Microsoft Power Platform automation supported internal maintenance and information workflows.
Evidence: Professional work at Modis / Baker Hughes, February 2022–March 2023. No public implementation or quantified performance results are presented here.

Remote sensing · Environmental data

Semantic segmentation and satellite data analysis for environmental measurement.

Problem: Estimating carbon stocks from satellite imagery requires structured geospatial data and image interpretation.
My contribution: At Plasive Technologies, I designed and trained CNN segmentation models, developed a carbon stock estimation method, and constructed a multidimensional satellite-image dataset.
System / method: Semantic segmentation of satellite imagery combined with remote sensing and data analytics.
Evidence: Professional work at Plasive Technologies in 2021. No public dataset, implementation, or validated performance figures are claimed here.