Ray / KubeRay
Open-source contributor on KubeRay's kubectl plugin — added S3, ephemeral storage, and TPU support for simpler Ray cluster deployment.
- Go
- Kubernetes
- Ray
- OSS
ML Systems Engineer
Eagle Lo
Building the systems behind learning machines.
Machine Learning Engineer II @ Samsara · CMU MSDS '26
Selected Work
What I've been building
Edge Signal Quality at Samsara
Led a full-stack ML project that improved asset-tracking location error by 38% — Spark on Databricks for billions of events, Ray Train + Ray Tune for XGBoost, Ray Serve + Go for production rollout. Converted to full-time MLE II on the ML Core Infra team.
- Ray
- Spark
- XGBoost
- Go
Twitter-Scale ETL & Infrastructure
Processed 1TB+ of Twitter data on Azure HDInsight; designed a denormalized MySQL schema sustaining 10K+ RPS; deployed microservices to EKS with Terraform + Helm + Aurora.
- Spark
- Kubernetes
- Terraform
- Azure
Latest Writing
Recent essays
Stream Processing: A Deep Dive into Kafka and Samza
An exploration of stream processing fundamentals, with a focus on Kafka and Samza — their use cases, processing guarantees, and the trade-offs between them.
GitHub Best Practices: Feature Development Workflow
A short guide to the workflow I use for new features — branching from a clean main, naming conventions, and the small habits that prevent merge pain.
Library Snapshot
What I'm taking in
Designing Data-Intensive Applications — Martin Kleppmann· In Rainbows — Radiohead· La La Land (2016)