Justin wang

I am an AI/ML engineer specializing in building production-grade generative AI and agent systems that deliver measurable business impact. I design end-to-end ML pipelines, incorporate safety guardrails, and operationalize models at scale across cloud-native infrastructures.

At Amazon, I architected multi-agent orchestration with Bedrock, Anthropic, and custom runtimes, improving beta delivery efficiency by 65% and standardizing model access with a Model Context Protocol. I designed RAG pipelines, vector search, and fine-tuning workflows to enable accurate multi-turn responses for advertiser workflows.

Previously at Oscar Health, I developed deep learning models for medical records and care-routing systems that improved condition detection accuracy and reduced episode costs significantly, while migrating real-time fraud detection to stream processing.

I prioritize robust monitoring, compliance, and cost-aware deployments—implementing guardrails, RBAC, observability, and GitOps—to reliably move AI projects from pilot to production and accelerate iteration cycles for cross-functional teams.