Building agent workflows and the data platform underneath them.
I'm Duyet — a Senior Data & AI Engineer. I focus on the foundational systems that make software work: data pipelines that scale, platforms that stay reliable, and AI agents that are actually useful. I build systems that are simple to operate and transparent about what they do — and I open-source most of what I build.
Things I've shipped
An autonomous agent skill designed to help build, run, and scale downstream AI agent systems.
Codex and Claude Code plugins for AI-powered development workflows.
One API for every AI model. Route traffic across providers with fallback, observability, and BYOK from a single OpenAI-compatible endpoint built on Cloudflare's edge.
ClickHouse monitoring with AI agent support for finding insights, monitoring clusters, and triaging activity
State store for AI agents — persistent memory and state management for agent workflows.
Model Context Protocol tools and integrations
Open Managed Agents — a self-hosted control plane on Cloudflare that drives agents on any sandbox backend from one UI and one API.
Query an LLM directly via DNS TXT records — `dig @llm.duyet.net 'explain quantum computing' TXT +short`
Import Claude Code usage data (ccusage) into ClickHouse for long-term analytics.
Analyze your Claude Code usage with DuckDB — query session stats, token costs, and model breakdown.
Rust programming language documentation in Vietnamese
URL shortener with analytics and custom domains
Collection of useful Helm Charts, well-tested with KinD and Kubeconform.
From the blog
Areas of Expertise
Data Engineering
8yrEnd-to-end pipelines from raw event streams to analytics-ready tables. ClickHouse as the primary OLAP layer — petabyte-scale ingestion, columnar query optimization, and materialized view patterns. Airflow for orchestration; Spark for heavy batch transforms.
AI Agent Engineering
4yrAgent frameworks, model routing, and evaluation infrastructure. AnyRouter for multi-provider fallback and BYOK routing. LangGraph and LlamaIndex for workflow orchestration. Usage analytics baked in from day one.
Cloud Infrastructure
6yrKubernetes clusters on AWS and GCP, Helm chart authoring, Terraform for IaC. Cloudflare Workers for edge compute — zero cold-start, global, cheap.
Backend & APIs
7yrREST and GraphQL services in Python, Rust, and TypeScript. Cloudflare D1 + KV for edge-native persistence. Durable Objects for stateful agent sessions.
Frontend & UI
5yrReact and TanStack Start with SSG. shadcn/ui + Tailwind as the component layer across 8 sibling apps. Performance-first: sub-second LCP, no hydration overhead for static routes.
DevOps & Observability
5yrGitHub Actions CI/CD with Turborepo caching. Structured logging into ClickHouse; MotherDuck for ad-hoc querying. OpenTelemetry traces for agent workflows.
Open Source
8yrPublic libraries, dashboards, and reference implementations. rust-tieng-viet, ClickHouse Monitor, ShareHTML, LLM Timeline — built and maintained openly.