Hi, I'm Obanijesu

Skills

• Python
• SQL
• Pandas
• Power BI
• Metabase
• Scikit-learn
• LangGraph
• LangChain
• OpenCV
• FastAPI
• Streamlit
• MongoDB
• Playwright
• Git

Projects

• Spe-adMy company

  Request a demo

  African fintechs lose money on FX spreads they can't see. Spe-ad is an AI copilot that shows treasury teams where, in plain English: the cheapest provider and route per corridor, the official vs. parallel naira gap, and a safe quote buffer based on real volatility.

• AI Research Agent

  Check repository

  A LangGraph agent that searches ArXiv, Wikipedia and the web to answer complex questions.

• Fraud Detection App

  Check repository

  Scores credit card transactions for fraud risk in real time with a trained classifier.

• Smart Attendance System

  Check repository

  Facial recognition on a Raspberry Pi that logs student attendance to Firebase using OpenCV.

• Diabetes Checker

  Check repository

  A screening app for hospitals with 85% accuracy, built with Scikit-learn and Streamlit.

Experience

• 99%transaction success rate after data-driven fixes
• ~40%faster data retrieval for three business teams
• 60%+less data prep time with an automated pipeline
• 86%accuracy on a customer churn model

• 2026 – present

  Founder at Spe-ad

• Nov 2025 – present

  Data and AI Automations Expert at StartButton Africa

  • Helped increasing the transaction success rate to 99%this was done by running daily data standups using Power BI dashboards as a visualization that exposed bottlenecks, then escalated to engineers to ship the fixes.
  • Worked hand in hand with engineers in increasing data retrieval rate on metabase for analytics, sales and account management with cron job updating the data every hour.
  • Built a WhatsApp AI agent that collects treasury rates and writes them into an API-connected Google Sheet, removing manual data entry.
  • Automated risk and transaction monitoring: Python and SQL (Metabase) jobs that alert Slack when anomalies occur during the day or transactions look fraudulent.
  • Forecast annual transaction volume with Prophet, Holt's linear model and regression, validated against Q4 actuals for executive review.
• Mar – Sept 2024

  Engineering Intern at TeleAfrica Communication

  • Built a MySQL pipeline that turns raw network logs into analysis-ready datasets, cutting prep time by over 60%.
  • Predicted customer churn with 86% accuracy (Scikit-learn), giving the retention team a list of at-risk customers to target.
  • Kept 99.9% uptime on critical network switch connections.
• Aug 2023 – Sept 2024

  Junior Data Analytics Officer at Silicon Tech Analytics

  • Built a heart disease prediction model with 86.89% accuracy (Random Forest), beating the logistic regression baseline by 7.89 points after tuning.
  • Found significant links across 12 clinical variables in hospital data, which shaped what the model learned from.