Olumide O.

Key Skills

Software

No software listed

Top Subject Matter

Medical imaging (CT scan) cancer classification

Customer churn/attrition prediction

AI itinerary generation using external data sources

Top Data Types

Text

Image

Document

Top Task Types

Diagnosis

Classification

Function Calling

Freelancer Overview

Respirix Web App (Thesis Project): AI model training for cancer type classification. Brings 3+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include TensorFlow, scikit-learn, and Gemini. Education includes Bachelor of Science, Babcock University (2025). AI-training focus includes data types such as Medical, DICOM, and Text and labeling workflows including Diagnosis, Classification, and Function Calling.

Labeling Experience

Treklyio – AI Travel Itinerary Planner (Personal Project): AI agent workflow automation

Function Calling

Developed an AI travel itinerary planner that ingests real-time information and generates itineraries in under a minute. The system used external search and places data and orchestrated multi-step tasks using Gemini-powered agents. This work supported automated workflow generation for downstream planning experiences. • Integrated SerpAPI and Google Places API for real-time data ingestion. • Orchestrated planning tasks using Gemini-powered agents. • Streamlined itinerary generation through automated pipelines. • Improved planning efficiency by ~60% through agent workflow automation.

2025 - Present

Bank Customer Attrition Prediction (Personal Project): AI model development for churn classification

TextClassification

Built an XGBoost model for bank customer attrition prediction using tabular customer records. The project included feature engineering to improve model stability and reduce prediction bias. A dashboard was deployed to visualize predictions and support proactive retention decisions. • Trained on 10k+ customer records for attrition risk classification. • Engineered features from demographics, transactions, and account activity. • Reduced prediction bias and improved stability via feature processing. • Deployed a real-time attrition dashboard for operational use.

2025 - Present

Respirix Web App (Thesis Project): AI model training for cancer type classification

Diagnosis

Trained a CNN on ~1k CT scans for cancer type classification, using medical imaging data as the primary dataset. The work involved preparing consistent model inputs through preprocessing pipelines to improve detection reliability. Performance was validated using ROC AUC and F1 metrics to evaluate classification quality. • Trained on CT scan imagery to support cancer classification. • Preprocessed images to enhance quality and ensure consistent inputs. • Automated image upload and deletion after inference to support dataset handling. • Validated results using ROC AUC (~0.89) and F1 (~0.92).

2024 - 2025

Education

Work History