Blessing E.

Key Skills

Software

OpenCV AI Kit (OAK)

Other

Top Subject Matter

Agricultural insect classification for e-commerce-style image categorization

Machine learning analysis of gene expression drivers and e-commerce customer behavior

Top Data Types

Image

Top Task Types

Classification

Freelancer Overview

Agricultural Insect Classification (image classification model via transfer learning). Brings 2+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include OpenCV AI Kit (OAK) and Other. Education includes Master of Science of Science, Aarhus University (2025) and Bachelor of Science, University of Debrecen (2023). AI-training focus includes data types such as Image, Computer Code, and Programming and labeling workflows including Classification, Computer Programming, and Coding.

Labeling Experience

Machine Learning and Gene Network Sales Analysis

Other

Conducted end-to-end sales and customer analysis using Python and PostgreSQL to model customer lifetime value and related business metrics. Performed data quality control and normalization on high-dimensional gene expression datasets in preparation for machine learning. Trained and evaluated machine learning models to identify obesity-related driver genes and to quantify their contribution to revenue. • Performed data quality control and normalization for modelling • Built value, retention, and churn analysis for customer behavior • Applied Lasso and Random Forest to obesity-related gene drivers • Used feature importance and model coefficients for gene discovery

2025 - 2026

Agricultural Insect Classification (image classification model via transfer learning)

OpenCV AI Kit (OAK)ImageClassification

Developed an agricultural insect image classification model using transfer learning with ResNet-50 to perform automated categorization from images. Applied data augmentation and regularisation to improve robustness and reduce overfitting during training. Evaluated the model against a baseline benchmark to support generalization performance across training data. • Used transfer learning with ResNet-50 as the core architecture • Applied data augmentation strategies to strengthen training data • Implemented regularisation techniques to reduce overfitting • Reported an achieved accuracy of 65% versus baseline

2024 - 2025

Education

Work History