Skip to content
▸ Decrypting profileKien NguyenSite Reliability Engineer · Photographer
Site Reliability Engineer

Kien Nguyen. SRE by day, photographer by night.

KubernetesGitOpsBlockchainObservabilityAI PlatformsPhotographyKubernetesGitOpsBlockchainObservabilityAI PlatformsPhotographyKubernetesGitOpsBlockchainObservabilityAI PlatformsPhotography
Paris--:--:--
Experience

From metro systemsto blockchain.

Eight years keeping safety-critical and financial infrastructure online, across four cities and two continents.

Ledger

Ledger

Senior Site Reliability Engineer · Ledger

2023 - Now

Paris, France · 48.86°N 2.35°E

Lead reliability for blockchain platforms while building Ledger's self-hosted AI platform for developer productivity.

  • Centralized MCP servers and model providers (AWS Bedrock, GitHub Copilot) behind LiteLLM as a single gateway.
  • Cost and usage governance across teams through Langfuse and LiteLLM.
  • Deployed 4 NVIDIA A10G GPU nodes on AWS Kubernetes served with vLLM: Mistral-7B for agents, Qwen3-8B for chat, Llama-Guard-3-8B for guardrails, BAAI/bge-m3 for embeddings.
  • Deployed Coder as the execution environment for an internal AI agent: employees open Jira tickets from Slack, the agent spawns its own Coder workspaces and handles the task end to end, working across multiple repos in parallel.
  • Migrated PostgreSQL to CloudNative-PG with GitOps; database observability via Percona Monitoring and Management (PMM), with dedicated Docker images for PMM and pgAdmin.
  • Operate blockchain nodes: Bitcoin, Ethereum and more.
KubernetesvLLMLiteLLMLangfuseCoderNVIDIA A10GPostgreSQLPMMBlockchain

Paris · 48.86°N 2.35°E

Capability radar

8 / 8

Education

FR

ECE Paris, Engineering School

Engineering Degree

Specialization

Computer ScienceBig DataCybersecurity

Paris, France

AI Platform

A self-hostedAI platform.

Designed and deployed an end-to-end AI platform for developer productivity at Ledger: RAG-powered chat, workflow automation, LLM observability, and on-demand dev environments powered by Coder.

0B+
Tokens processed
0K+
Requests / week
0%
Success rate
0+
Services
Access
Interface
Automation
Intelligence
Data

Component

LiteLLM

LLM gateway

LiteLLM gateway unifying access to every model behind one OpenAI-compatible API: prompt routing, load balancing, spend tracking and fallbacks across providers.

KubernetesArgoCDHelmMCPRAGpgvector

Request flow

User
Okta
LibreChat
LiteLLM
RAG
pgvector
Langfuse

Architecture & design

Designed the full platform: LLM routing, RAG pipeline, observability, and evaluated each component for enterprise readiness.

GitOps deployment

Shipped all 10+ services on Kubernetes via ArgoCD + Helm, Terraform IaC, zero-downtime progressive rollouts.

Observability & ops

Built the observability layer: Langfuse for LLM tracing, Datadog for infra, and custom dashboards.

Photography

The other half:
chasing light.

View all albums
A dramatic Arctic archipelago where mountains meet the sea.
Aurora

Lofoten, Norway

A dramatic Arctic archipelago where mountains meet the sea.

01 / 04
Contact

Let's build somethingreliable, or chasethe next light.

Whether it's infrastructure, AI platforms or a remote summit at golden hour. The best way to reach me is a LinkedIn DM.

Off the clock

Photography

Sony A7IV

Piano

9 years

Cycling

Road & gravel

Tennis

Competitor

Got a quick question?

The assistant in the bottom-right corner knows my work, stack and projects. Ask it anything.

KNSite Reliability Engineer
© 2026 Kien NguyenNext.js · Tailwind · Kubernetes