Jack Lin
Principal Software Engineer at 台灣積體電路製造股份有限公司
Taipei, Taiwan
Summary
🤩
Rockstar🎓
Top SchoolJack Lin is a Principal Software Engineer based in Taipei with 11 years of hands-on experience building cloud-native infrastructure and MLOps platforms. He has driven scalable ML training and inference pipelines—at Appier he used Kafka and Argo to train hundreds of models and infer 100 million users in 30 minutes—and contributed core operator features to Kubeflow’s training-operator used for distributed ML on Kubernetes. At SUSE and InfuseAI he focused on reliability and security of production systems, implementing Kubernetes operators, RBAC/Istio integrations, and improving CI/CD and scheduling for high-availability workloads. His open-source work includes key reliability fixes and enhancements to kubeflow/trainer such as PDB owner references, ActiveDeadlineSeconds, BackoffLimit, and gang-scheduling integration. Comfortable bridging research and production, he pairs a Master’s in Computer Science with practical experience mentoring GSoC projects and shipping platform-level tooling. A detail-oriented engineer, he often surfaces operational constraints early, turning them into durable automation and testing improvements.
11 years of coding experience
6 years of employment as a software developer
Master's degree Computer Science, Master's degree Computer Science at Technische Universität DresdenMaster's degree Computer Science, Master's degree Computer Science at National Tsing Hua University
Github Skills (10)
kubeflow10
kubernetes10
distribute10
devops10
cicd10
kubernetes-pods10
dockers9
docker9
python8
machine-learning6
Programming languages (12)
Github contributions (5)
kubeflow/trainer
Feb 2018 - Jan 2022
Distributed ML Training and Fine-Tuning on Kubernetes
Role in this project:
DevOps Engineer
DevOps EngineerContributions:10 reviews, 19 commits, 25 PRs in 3 years 11 months
Contributions summary:Jack primarily contributed to the infrastructure and operational aspects of the Kubeflow trainer project. Their work involved fixing bugs related to the creation and deletion of PodDisruptionBudgets (PDBs), which are critical for high availability during training. They also implemented features such as adding owner references to PDBs and integrating ActiveDeadlineSeconds and BackoffLimit for job management. Furthermore, they addressed issues in the CI/CD pipeline and modified the project to align with Kubeflow common practices.
xgboostkubernetesmachine-learningtrainingai
ChanYiLin/tf-operator-Dragon
Apr 2018 - Jan 2019
Extends Kubeflow/tf-operator to enable gang-scheduling and auto-scaling.
Contributions:38 commits, 27 pushes, 5 branches in 9 months
auto-scalingschedulingoperatorkubeflowkubernetes
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.