Work Experience

AI/ML Solutions

• Accelerated claims processing by 40% through custom OCR and object detection models integrated into SnapInsure app
• Increased claims validation accuracy by 45% via RAG model using FAISS and LangChain for real-time document analysis
• Built electronic signature model integrated into core application enabling seamless document signing with YoloV8
• Developed claim cause-to-loss validation system using damage detection models to verify reported damages
• Built cost estimator for damaged car parts with instant web scraping for price comparison
• Implemented Ghana Card OCR system for automated identity verification with paddleOCR
• Developed object tracking model for vehicle inspection to prevent fraud during claim uploads
• Built ASR models using OpenAI Whisper and PyAnnote for speech diarization and multi-language translation
• Developed trash classifier model using YOLOv11 for deployment on robotic arm systems
• Built face recognition-based attendance system with anti-spoofing capabilities

AI Agent Development

• Built fully autonomous AI Claims Agent with conversational chat interface and Claude API integration
• Developed RAG engine with Claude Vision OCR for document upload and intelligent Q&A
• Implemented comprehensive fraud detection system with image metadata analysis and AI-generated image detection
• Built intelligent web scraping system for real-time market price lookup
• Developed adaptive learning engine that improves agent performance through feedback collection

Software Development

• Developed comprehensive investment management module for Phrontlyne ERP system
• Built memo management module for internal document workflows
• Automated critical business processes, improving operational efficiency by 50%
• Architected interactive Reinsurance Dashboard using Metabase

• Built end-to-end ML system for Ghana Enterprise Agency to analyze ~109,000 female applicant profiles across 4 Mastercard-funded programs
• Developed ML eligibility predictor using Logistic Regression, Random Forest, and Gradient Boosting models
• Created application screening dashboard with priority scoring (0-100) for efficient high-volume processing
• Built viability and resilience prediction models analyzing 28+ financial features

• Built Ghana's AI-powered environmental monitoring platform for detecting illegal mining, deforestation, and pollution
• Developed computer vision models for real-time threat detection analyzing satellite imagery
• Implemented backend infrastructure with authentication system and real-time data processing pipelines
• Integrated platform with Ghana Police Service, EPA, and local agencies for coordinated environmental protection

• Fine-tuned Mistral model for AI mentorship platform, improving interaction quality by 15%
• Developed mentor-mentee matching algorithm using cosine similarity, achieving 30% better matching accuracy

• Standardized and documented data protocols to maintain high-quality, consistent data across diverse sources
• Managed and integrated large, multi-source datasets for data-driven decision-making
• Created impactful data visualizations and conducted advanced statistical analyses using Matplotlib

• Developed diabetes prediction models achieving 85% accuracy using diverse patient attributes
• Created interactive Tableau dashboards, improving stakeholder engagement by 25%