About

Hi, I’m The Anh Nguyen (@ntheanh201), based in Hanoi, Vietnam. I lead a team building GPU infrastructure and LLM serving on Kubernetes, and I speak at community events about what we learn along the way.
Now
I’m tech lead and solutions architect for the Viettel AI Platform at Viettel Networks, a multi-tenant GPU platform and a Certified Kubernetes AI Platform under CNCF’s AI Conformance program. That includes Token Factory, which serves LLMs at scale.
- Token Factory: LLM inference at scale with NVIDIA Dynamo, serving multiple LLM families
- GPU fleet: NVIDIA H200 clusters on NVIDIA AI Enterprise and Base Command Manager (BCM)
- GPU and fabric: GPU Operator and Network Operator, InfiniBand with GPUDirect RDMA, multi-node NCCL, and HAMi fractional-GPU sharing for multi-tenancy
- Upstream: contributing to NVIDIA DSX OS components: KAI Scheduler, Run:ai, Workload Map, AI Cluster Runtime (AICR) and Cluster Readiness Engine (NVCRE)
Recognition
- CNCF Ambassador, Cloud Native Computing Foundation
- Kubestronaut: CKA, CKAD, CKS, KCNA, KCSA
- NVIDIA-Certified Professional: AI Operations (NCP-AIO), AI Infrastructure (NCP-AII), AI Networking (NCP-AIN)
- Claude Certified Architect – Foundations, Anthropic
- The Shubhra Kar Linux Foundation Training (LiFT) Scholarship, 2025 recipient
- Outstanding Individual, Viettel’s global IT line, 2025, Viettel Group
- Two Second Prizes, VTNet 2025 Innovation Journey, with my team, Viettel Networks
- Certified Kubernetes AI Platform, CNCF Kubernetes AI Conformance (v1.35), for the Viettel AI Platform (team)
- Silver, 2026 Globee Awards for Artificial Intelligence, GPU-as-a-Service (GPUaaS) & Compute Orchestration, for the Viettel AI Platform (team award)
- Turning GPU-hours into tokens, my case study on serving multiple LLM families with NVIDIA Dynamo, is listed in NVIDIA Dynamo External Publications
See all certifications, including NVIDIA NCP-AII and NCP-AIN, LFCS and the Nebius AI certifications.
Open source
| Project | What I do |
|---|---|
| kai-scheduler/KAI-Scheduler | NVIDIA DSX OS: Kubernetes-native scheduler for AI workloads |
| dsx-ai-factory/run-ai | NVIDIA DSX OS: Run:ai AI workload orchestration platform |
| dsx-ai-factory/workload-map | NVIDIA DSX OS: maps framework CRDs into a generic workload structure |
| NVIDIA/aicr | NVIDIA DSX OS: AI Cluster Runtime (AICR), validated, version-locked recipes for GPU Kubernetes clusters |
| NVIDIA/cluster-readiness-engine | NVIDIA DSX OS: Cluster Readiness Engine (NVCRE), certifies GPU clusters under real distributed load before production |
| kubernetes/website | Kubernetes org member, SIG Docs Vietnamese localization approver |
| cncf/glossary | Vietnamese localization lead for the Cloud Native Glossary |
| Project-HAMi/HAMi | Org member; fractional GPU sharing for Kubernetes |
| kubeflow/kubeflow | AI/ML platform on Kubernetes |
| apache/trafficcontrol | CDN; upstreamed the Grafana Scenes dashboards |
Talks
I speak at community events in Vietnam, from Cloud Native Hanoi meetups to KCD & OpenInfra Days. Each talk below links to a recap on this site.
- Turning GPU-Hours into Tokens, and Tokens into a Service: Agent Router and NVIDIA Dynamo, VOI & CNVN x AWS FCAJ Meetup: Cloud & AI Infrastructure, Hanoi, 2026 · Recap
- From Project to Production: HAMi and Viettel Cloud, KCD & OpenInfra Days Vietnam 2026 · Recap
- Auditable Autonomy: Engineering With AI Teammates That Leave a Trace, The Future of Software Engineering in the Agentic AI era · Recap
- From Vietnam to the World: Contributing to Kubernetes and earning the Linux Foundation Scholarship, #cTENcf Birthday Bash Hanoi · Recap
- AI GÕ CỬA — Doanh nghiệp Việt đã sẵn sàng?, VIETTELDX TALKS #8 · Recap
- HAMi: Fractional GPU for Efficient AI in Cloud Native, OpenInfra & Cloud Native Day Vietnam 2025 · Recap
- Fractional GPU for AI: Open Source Solutions in Cloud Native, Cloud Native Hanoi May Meetup · Recap
Competitions
As tech lead of the AI Platform, I’m responsible for the GPU platform these competitions run on:
- Viettel AI Race 2025 and 2026: the environments every team trains and submits on (Hackathon Day 2025)
- Vietnam AI Open Hackathon 2026, with NVIDIA: on the technical support team; the cluster passed all 14 checks in NVIDIA’s Cluster Readiness Checklist (post)
Projects
- trafficcontrol-grafana-scenes: dynamic Grafana dashboards for Apache Traffic Control built on Scenes, upstreamed in trafficcontrol#7927
- k8s-auto-healing: a Go controller that runs periodic health checks and auto-remediates known failures (e.g. Zalando Postgres replication lag) from Prometheus alerts
- react-user-feedback: a drop-in feedback widget for React apps