Babatunde Oshodi

Supported enterprise GPT model evaluation and fine-tuning across DevOps, medical AI, and diagnostic domains. Authored and optimized prompts, graded model responses, and ensured 95%+ accuracy on deliverables. Implemented annotation guidelines and QA processes for consistent labeling and evaluation. Provided prompt-engineering expertise to improve contextual reliability of LLM responses.

Designed and deployed a HIPAA-compliant AI cancer-screening platform using cloud-native ML architecture. Migrated healthcare SaaS workloads to AWS using Terraform, Kubernetes (EKS), and secure VPC/IAM networking. Processed and analyzed 133+ GB of CT-scan data, boosting inference pipeline performance by 25%. Built Kubernetes-based inference pipelines with GPU autoscaling and integrated Prometheus/Grafana dashboards. Engineered ETL pipelines, model-tuning workflows, and RAG prototypes for clinical decision support. Integrated Elasticsearch for real-time log ingestion, monitoring, and system performance tuning. Led incident response and troubleshooting for ML/ETL pipelines, improving system uptime and reliability. Operated in Agile/Scrum frameworks, driving sprint planning, stand-ups, and retrospectives. Shipped ML/ETL logs to ELK and Splunk; built Kibana/Splunk dashboards and log-based alerts for inference latency, GPU utilization, and data quality.

Designed and delivered the Window Genius AI CRM integrating AI chatbots, Stripe SaaS billing, and Facebook Webhooks. Migrated production from Heroku to AWS ECS Fargate, achieving improved scalability, security, and uptime. Automated container deployments via GitHub Actions → Amazon ECR → ECS Fargate, and configured ALB for high availability. Enhanced chatbot architecture with AWS Lex, AppSync, and API Gateway for real-time lead capture and scheduling. Improved lead response time by 60% using Webhook automation and Stripe API integration. Enabled real-time analytics with Grafana, CloudWatch, and PostgreSQL, boosting operational visibility.