CV

Machine Learning Engineer (Intern)

Stripe

• Developed and deployed ML models for secret key leak detection, cutting detection latency from 4.5 days to under 1 hour and preventing multi-million-dollar annual fraud losses. • Engineered large-scale feature pipelines in Spark/Databricks (500M+ records across Iceberg tables), integrated with Shepherd and Airflow for production-ready workflows. • Expanded detection coverage by onboarding new API-call signals (OAuth, Connect), boosting recall on confirmed leaks to 94% while ensuring system observability with custom dashboards.

Graduate Teaching Assistant

University of Georgia

• Collaborated with Prof. Shelby Funk and Dr. Michael Cotterell in delivering course content for Theory of Computing and Human-Computer Interaction. • Evaluated and graded assignments and exams with adherence to university grading standards, providing timely and constructive feedback.

Software Developer (Intern)

Orbital Schools

• Spearheaded the development of dynamic websites utilizing the LAMP stack, enhancing client digital presence with Linux, PHP 5, CSS, and MySQL. • Executed seamless integration of payment processing capabilities by embedding PayStack and other APIs into Orbital Schools' web platform.

PhD

University of Georgia

Computer Science

• Machine Learning, Algorithms, Software Engineering, Data Mining, Trustworthy Machine Learning, Databases, Advanced Robotics, Human-Computer Interaction and Advanced Representation Learning

42 Silicon Valley

Software Engineering

BSc

University of Ilorin

Computer Science

Adaptive Quantile Guidance in Diffusion Models: Multi-Dataset Learning for Pandemic Time Series. ICMLA 2025 (Accepted)

• Developed AQDiff, a diffusion-based framework for epidemiological forecasting, leveraging multi-dataset pre-training (CDC influenza, COVID-19, RSV) and adaptive quantile guidance to dynamically adjust prediction intervals. Achieved 73.3% lower MAE than state-of-the-art models (CSDI, PatchTST) across pandemic scenarios. • Introduced real-time error feedback via exponentially weighted residual statistics, enhancing volatility adaptation with <10% computational overhead. Demonstrated robustness in sparse data methods and statistically significant improvements during outbreak surges.

EpiCast

Built a mobile-first disease surveillance platform for West Africa (ECOWAS) combining fine-tuned medical LLMs with audio AI for syndromic monitoring and cough-based respiratory illness detection. Fine-tuned and quantized MedGemma 4B to GGUF format for on-device inference. Integrated MedGemma SigLIP for clinical image triage, HeAR audio models for cough classification, and a multilingual NLP pipeline trained on 3,400+ corpus-grounded examples across French, Hausa, and Yoruba.

• MedGemma 4B/27B
• MedGemma SigLIP
• Google HeAR
• FastAPI
• RunPod Serverless

Omnia

Engineered a full-stack generative AI pipeline that transforms couples' narratives into personalized animated short films using multi-modal large language models for story understanding, scene decomposition, and character-consistent image generation. Implemented a two-stage generation architecture with LLM-driven scene synthesis and image-to-video animation.

• Stable Diffusion
• AnimateDiff
• React
• Node.js/Express
• Cloudflare R2

Pincel (Diagramify)

Developed an AI-powered tool that automatically generates publication-quality architecture diagrams from research papers and codebases, reducing manual diagram creation time from 4-8 hours to minutes. Built a two-stage pipeline with PDF parsing and TikZ/SVG diagram generation trained on ML conference papers.

• NLP/LLM
• TikZ/SVG
• React
• NeurIPS/ICML/ICLR styles

OmniAsset

Built a React Native mobile app that tracks six asset classes (stocks, crypto, real estate, vehicles, precious metals, cash) with live market pricing, unified net worth calculation, and AI-powered portfolio risk analysis. Integrated a financial LLM agent for personalized portfolio critique and a FIRE calculator with interactive wealth projections.

• React Native
• Financial LLM Agent
• FIRE Calculator
• Mobile App

Factor-Social

A university social platform built to help students connect and integrate into campus life. Creates a central, university-branded hub where students can discover clubs, form groups, find events, and interact with classmates and campus resources.

• Social Platform
• Flutter
• Mobile App

Prompt Fusion in Stable Diffusion

Optimized image generation by consolidating positive and negative prompts into a single efficient pass. Utilized LPIPS and Siamese networks for evaluation, and BERT for prompt similarity, improving optimization time and image-prompt alignment.

• Generative Models
• Prompt Fusion
• Stable Diffusion

HumorAI: Enhancing LLM for Implicit Humor and Sentence Analysis

Engineered an advanced LLM model for sentiment analysis integrating humor recognition using contrastive loss on SBERT and BERT. Achieved 15% improvement in accuracy on chatbot and Twitter similarity tasks over existing models.

• LLMs
• Sentiment Analysis
• NLP

StockCluster: Enhancing Investment Portfolios

Devised a clustering algorithm for stock market analysis, achieving a 6.89% ROI increase by optimizing portfolio diversification.

• Clustering
• Stock Diversification
• ML

Risk-Aware Task Allocation in Multi-Robot System

Developed a risk-aware multi-robot task allocation model factoring in task dependencies and robot characteristics, increasing operational efficiency by 40% and improving task completion rates.

• Multi-Robot Systems
• Task Allocation
• Optimization