Andrew Osborn

Led end-to-end model inference dashboards with a Next.js/React frontend and FastAPI/Python backend, scaling to 500K+ daily inference requests across global edge deployments with a responsive UI and high-throughput API design. Architected and integrated RAG pipelines and agentic AI workflows into the Hugging Face Hub core platform using LangChain, LlamaIndex, and OpenAI/Anthropic APIs — enabling 2M+ developers to build production-grade LLM applications with 65% improved response accuracy and reduced hallucinations. Designed and optimized data pipelines and vector database solutions using PostgreSQL + pgvector and DynamoDB, powering AI-driven semantic search and recommendation engines; improved query throughput and reduced end-to-end latency through batching, indexing, and async ingestion strategies. Cut query latency by 45% by deploying optimized vector embeddings with Redis caching layers and async task queues (Kafka/RabbitMQ) for inference request fan-out, ensuring consistent perfor

Built end-to-end model inference dashboards with a Next.js/React frontend and FastAPI/Python backend, scaling to 500K+ daily inference requests across global edge deployments with a responsive UI and high-throughput API design. Architected and integrated RAG pipelines and agentic AI workflows into the core platform using LangChain, LlamaIndex, and OpenAI/Anthropic APIs, enabling 2M+ developers to build production-grade LLM applications with improved accuracy and reduced hallucinations. Designed and optimized data pipelines and vector database solutions (PostgreSQL + pgvector, DynamoDB) powering AI-driven semantic search and recommendations; improved query throughput and latency via batching, indexing, and async ingestion. Cut query latency by 45% by deploying optimized vector embeddings with Redis caching layers and async task queues for inference fan-out. Led migration to domain-driven microservices on AWS EKS + Bedrock, increasing deployment frequency and ensuring fault isolation. Ow

Spearheaded full-stack development of MinuteClinic Digital Tools (Clinic Scheduling & Virtual Care), a scalable platform connecting patients with in-clinic and virtual healthcare, driving a 25% increase in online appointments and significantly improving patient engagement. Integrated AI/ML capabilities for predictive patient engagement and personalized healthcare recommendations using Python, TensorFlow, and Databricks; architected backend microservices using Java, Spring Boot, Node.js, REST/GraphQL, and integrated with Epic EHR, Redis, RabbitMQ/Kafka, ensuring secure real-time patient data access at scale. Designed robust data pipelines using PostgreSQL, DynamoDB, and ETL workflows, improving query performance by 30% under high-load conditions and enabling reliable analytics and reporting. Reduced system latency by 40% through asynchronous processing, intelligent caching strategies, and message queue architecture, ensuring scalability during peak traffic periods. Led cloud-native arch