Open to opportunities

Nathan Leung

@nathanleung

Message

Staff Full-Stack ML Engineer specializing in GenAI platforms, MLOps, and scalable production AI systems.

United States

Message

What I'm looking for

I seek hands-on engineering roles building scalable GenAI/MLOps platforms where I can lead architecture, drive CI/CD and IaC, and enable data science self-service.

I am a Staff Full-Stack ML Engineer with 14+ years building production systems at scale, from ML training and deployment infrastructure to GenAI platforms. I design and deliver end-to-end solutions including RAG pipelines, LLM fine-tuning, multi-tenant AI SaaS, and event-driven services on AWS.

At AllCloud I architected a multi-tenant GenAI platform using Bedrock, OpenSearch Serverless, EKS, and Terraform, and led fine-tuning and deployment of custom models. Previously at Twitch and Amazon I built large-scale ML training pipelines, model registries, feature pipelines, low-latency model serving, and production ad and payment systems.

I bring strong hands-on expertise across cloud infrastructure, Kubernetes, IaC, SageMaker, PyTorch/TensorFlow, and distributed systems, paired with leadership in architecture, GitOps-driven delivery, and enabling data scientists to self-serve ML workflows.

Experience

Work history, roles, and key accomplishments

Current

Staff Full-Stack ML Engineer

Current

AllCloud

Sep 2023 - Present (2 years 9 months)

Architected a multi-tenant GenAI SaaS platform and end-to-end RAG pipelines using Amazon Bedrock and OpenSearch Serverless, enabling contextual recommendations and tenant-isolated scalable inference. Delivered fine-tuned LLMs, Terraform IaC, and GitOps CI/CD to production, reducing operational complexity and standardizing LLM integrations.

OpenSearch RAG EKS Terraform Step Functions S3 KEDA GitOps

Senior ML Platform Engineer

Twitch

Jul 2021 - Aug 2023 (2 years 1 month)

Built end-to-end SageMaker training and deployment pipelines, model registry, and feature pipelines ingesting billions of events, reducing model deployment time from days to hours and enabling reproducible, scalable ad-serving ML systems. Implemented real-time serving, canary rollouts, and A/B testing for production models.

Sagemaker PyTorch TensorFlow Kinesis Model Registry A B Testing Real Time Serving

Senior Software Engineer

Twitch

Apr 2016 - Jun 2021 (5 years 2 months)

Designed and launched interactive ad formats and the Bounty Board marketplace using Go microservices and DynamoDB, increasing ad engagement and enabling thousands of brand-streamer partnerships; led migrations from monolith to microservices and built A/B experimentation infrastructure.

Go DynamoDB Kinesis Microservices A B Testing React RPC Event Driven Systems

Software Development Engineer II

Amazon

Jul 2011 - Apr 2016 (4 years 9 months)

Architected client-side and backend flows for Amazon Appstore IAP, built resilient transaction handling and server-side receipt verification, and developed search/ranking and promotional systems that supported large-scale app discovery and high-volume downloads.

Java Android SDK DynamoDB SQS REST APIs Distributed Systems

Software Development Engineer Intern

Amazon

May 2010 - Aug 2010 (3 months)

Built automated testing and load-testing frameworks for the Amazon Appstore backend, enabling CI/regression testing and validating system capacity prior to public launch. Produced realistic test data generation utilities to support integration testing.

Test Automation Load Testing Data Generation CI Systems Appstore QA Performance Testing