Description

AVAILABLE FOR NEW PROJECTS STARTING 01.09.2025.

Former astrophysicist and ESA/NASA Euclid mission contributor specialising in high-precision computer vision for mission-critical applications. I develop advanced Computer Vision and ML solutions where failure is not an option - from space-based imaging to vision-based navigation in GPS-denied environments. My expertise includes deep learning (YOLO, ResNet), 3D modelling, and statistical inference, with proven leadership scaling AI teams to 90%+ forecast accuracy under resource constraints. Currently leading R&D on autonomous vision systems via the early-stage VESTA project. I deliver custom Computer Vision models, prototype validation and end-to-end AI development for startups and enterprises demanding exceptional precision and reliability.

Full CV/resume available on request.

Industry field of expertise

Languages

English
Native or bilingual

Workplace preferences

Remote only

Primarily works remotely

VESTA
Founder (Pre-incorporation)
DIGITAL AND IT
May 2025 - Today (1 year and 1 month)
- Developing a vision-only perception layer for autonomous navigation in GPS-denied environments, targeting robotics and space applications.
- Built crater segmentation and feature extraction models for Martian terrain
- Defined technical roadmap from concept to early validation using planetary datasets
- Outlined commercial strategy for Earth pilots and planetary mobility use cases
- Prepared business and incubation proposals; pitched at ESA Business Incubation Centre (June 2025)
Vision Systems for Autonomy (Independent Project)
Independent Computer Vision Researcher
DIGITAL AND IT
November 2024 - April 2025 (5 months)
Explored Computer Vision approaches for terrain understanding in unstructured, GPS-denied environments.
- Conducted terrain segmentation and image analysis on Mars HiRISE data under low-texture and low-light conditions, using Python, OpenCV, ResNet, Detectron2 and Colab Pro
- Identified limitations in standard CV methods for terrain perception, shaping the early-stage design of the VESTA system
Greentech AI Startup
Senior Machine Learning Scientist & Technology Lead
DIGITAL AND IT
February 2022 - August 2024 (2 years and 6 months)
Brandenburg, Germany
Streamlining indoor farming operations by delivering precise CV-based crop yield forecasts and decision-making tools.
- Hyperparameter tuning of models: Significantly improved fruit detection accuracy via YOLO CNN models using the PyTorch framework and Weights & Biases MLOps platform. Projections suggest these enhancements would achieve over 95% accuracy in tomato yield predictions, thereby meeting the MVP goal.
- Developed Key Training Data Tool: First crop growth simulator in Python, addressing initial data scarcity and enabling the pre-training of ML models and the design of CV-based plant experiments.
- Developed Time-series model (NeuralProphet): Trained on climate & harvest data (88% accuracy pre-CV integration), delivering rapid results for investors and informing subsequent deep learning improvements.