Babatunde O.

Key Skills

Software

No software listed

Top Subject Matter

Biomedical ML and drug discovery (cancer, targets, RNA-seq/Ribo-seq)

Clinical NLP text classification

Computer vision for plant disease classification

Top Data Types

Text

Image

Top Task Types

Fine-tuning

Freelancer Overview

Graduate Research Intern (Nutrition and Industry Lab, Biochemistry, University of Ibadan). Brings 4+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Python, Streamlit, and CSS. Education includes Bachelor of Science, University of Ibadan (2024). AI-training focus includes data types such as Text, Medical, and DICOM and labeling workflows including Computer Programming, Coding, and Fine-tuning.

Labeling Experience

Independent Research Project: Plant Disease Classification Using Deep Learning

ImageFine-tuning

Fine-tuned a CNN image classifier using a multi-class plant disease dataset to produce labeled predictions across many disease categories. Used transfer learning from an ImageNet-pretrained ResNet18 and supplemented training with standard augmentation strategies. Applied early stopping and baseline comparisons to validate gains from transfer learning. • Trained ResNet18 on PlantVillage images with 38 disease classes across 14 crops • Implemented a custom CNN baseline and demonstrated performance improvement from transfer learning • Added data augmentation (rotations, flips, color jitter) to improve generalization • Used early stopping to reduce overfitting and support robust crop monitoring models

2025 - Present

Independent Research Project: Medical Text Classification with Transformers

Fine-tuning

Fine-tuned transformer-based NLP models on a labeled medical abstracts dataset to classify texts into clinical categories. Implemented robust preprocessing and training procedures to ensure stable and reproducible fine-tuning. Evaluated domain-pretrained biomedical transformers against general-purpose baselines using validation accuracy. • Fine-tuned BERT-family models including BioBERT and BioClinicalBERT for five clinical categories • Built stratified splitting, label normalization, early stopping, and checkpointing into training pipelines • Compared biomedical domain-pretrained models to general-purpose models to quantify performance differences • Supported clinical NLP training workflows for medical literature mining and decision support

2025 - Present

Graduate Research Intern (Nutrition and Industry Lab, Biochemistry, University of Ibadan)

Text

Built an AI-enabled pipeline that supports computational pharmacology and downstream model training using biomedical datasets, including drug–cell line data and biochemical databases. Developed preprocessing, feature engineering, and model optimization workflows to enable predictive ranking and classification tasks. Supported validation using docking and simulation-based checks rather than manual annotation. • Designed automated RNA-seq and Ribo-seq differential expression analysis workflows for experimental comparisons • Built ML/DL drug response and therapeutic target screening models from structured biomedical data sources • Implemented a multi-parameter optimization web tool to rank compounds using configurable IC50 and ADMET-like inputs • Used reproducible training pipeline practices (splitting, checkpoints, early stopping) for model stability in clinical NLP tasks

2023 - Present

Education

Work History