About Kevin
I bridge the gap between complex research and functional software.
I bridge the gap between complex research and functional software. While many can build models, I specialise in pioneering data solutions in highly specialised domains -from Biotech and AgTech to FinTech.
I don't just apply algorithms; I build the scripts, tools, and cloud-integrated engines required to turn an idea into a product. My background is rooted in deep technical research, backed by a patent and peer-reviewed publications.
How I add value to your project
- Custom ML Architecture: I design and deploy end-to-end models, ensuring they integrate seamlessly into your environment.
- Automated Intelligence (NLP): I build text-mining tools to extract entities and patterns from unstructured data and scientific publications.
- Computer Vision & Monitoring: I develop software for visual data analysis, such as phenotyping, drought detection, and geometric pattern recognition.
- Feasibility Testing (PoC): I run high-speed experiments to prove a technology's viability before you commit to full-scale investment.
- Technical Leadership: I lead international teams to align AI constraints with business goals and ROI.
Proven Cross-Domain Success
- Life Sciences: Developed computational methods for HIV genomics and protein crystal structure analysis.
- AgTech: Designed instruments and software for crop stress monitoring and seed germination.
- Text Mining: Built systems for automated protein entity recognition from biomedical literature.
- Risk Systems: Architected cloud-based engines for automated decisions-making and predictive analytics.
Technical Stack
- Languages: R (Expert), Shell Scripting, SQL, Python, Java.
- AI/ML: Machine Learning, NLP, Computer Vision.
- Tools: MLOps.
Looking for a technical partner to turn your data into a working product? Let's schedule a brief intro call.
English
Native or bilingual
German
Native or bilingual
Remote only
Primarily works remotely
Experience
- INFORM GmbH - Optimization SoftwareTeam Lead Shared Consulting ServicesBANKING AND INSURANCEJanuary 2023 - Today (3 years and 6 months)Aachen, Germany
- Smart Claim Management: To solve the problem of slow and manual insurance workflows, I developed an automated AI strategy by aligning user needs with technical constraints in Azure, using explainable models (tree-based models combined with TreeInterpreter algorithm) to achieve faster processing, and better business-tech alignment.
- Privacy-Preserving Fraud Detection: To address strict data privacy laws blocking model training, I created a decentralised detection system using a propitiatory software, H2O.ai, xgboost, R, and PMML to train models without moving raw data, achieving 100% privacy-compliant fraud detection.
- Real-Time Risk Infrastructure: To address the lack of risk model-assisted real-time processing, I architected a live decision engine where data were extracted via cloud-based tools and processed using proprietary feature engineering software. I developed and containerised the machine learning trainings within Docker, ultimately deploying them as PMML files through proprietary software to enable real-time, automated fraud decisions at scale.
- INFORM GmbH - Optimization SoftwareData Scientist / IT ConsultantBANKING AND INSURANCEApril 2018 - December 2022 (4 years and 8 months)Aachen, Germany
- Fraud Detection Proof-of-Concept: To reduce the high investment risk of new ML projects, I provided low-cost pilot validation by testing anomaly detection on banking data using Isolation Forest, G means clustering, Random Forest, Gradient Boosting Machine, XGBoost, and Explainable Boosting Machine, which confirmed ROI before a full-scale rollout.
- Customer Defaulter Prediction: To prevent financial losses from unpaid bills, I built an early-warning system by training classification algorithms in R on customer behaviour data, identifying potential defaults before they occurred.
- LemnaTecApplication ScientistBIOTECHJanuary 2014 - March 2018 (4 years and 2 months)Aachen, Germany
- Automated Plant Phenotyping: To replace subjective and slow manual monitoring, I created a visual monitoring instrument using a proprietary product solution based on OpenCV for data extraction, providing researchers with high-accuracy, objective data.
- Crop Stress Detection: To help researchers identify early signs of crop failure, I developed a visual extraction workflow using a proprietary production solution based on OpenCV to quantify phenotype dynamics, resulting in objective data for cereal resilience studies.
Recommendations
Be the first to recommend Kevin
Help this freelancer shine by sharing your experience working together.
These freelancer profiles also match your criteria
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Education
- PhD, Biological SciencesUniversity of Cambridge2010Thesis: Functional annotation of predicted active sites - PDB and literature mining. Research domain: Biomedical Literature Mining, Data Integration, Protein Structure Data Mining, and Bioinformatics.
- MSc BioinformaticsCranfield University2003Thesis: Novel protein structure prediction method - utilisation of a peptide conformation library derived from non-parametric statistical analysis. Courses: Bioinformatics, statistics, macromolecular modelling and analytical science.