Operator // online

Chinnakrit Sripan

AI Automation & Full-stack Systems Builder

I build practical AI automation and full-stack systems that connect web apps, internal APIs, server operations, and real notification channels. Recent work includes a LINE-connected AI assistant, Codex-to-Hermes DevOps workflows, and a read-only knowledge assistant from a sanitized Obsidian vault.

Open transmissionView featured systems

Status

Building production AI workflows

Location

Thailand

Command modes

BuildAutomateOperateOptimize
// 02Signal // metrics

Systems shipped, not slides made.

A snapshot of what is currently live, monitored, and iterated on across personal infrastructure.

Production systems

12+

Live across web, ops, and assistants

Automation workflows

20+

Connecting APIs, alerts, and operators

Targeted uptime

99.x%

Health-checked and recovery-ready

// 03Surface // capability

Where the operator console connects.

Six surfaces that recur across the systems I build — each one designed to be shipped, observed, and handed off.

01

AI assistant workflows

Messaging assistants, prompt flows, knowledge context, and operator handoff.

02

Internal tools

Admin dashboards, review flows, content tools, and daily operating surfaces.

03

DevOps automation

Health checks, alerts, reports, event logs, and deployment-ready runbooks.

04

Knowledge systems

Sanitized read-only exports, RAG-style patterns, and wiki retrieval flows.

05

Full-stack systems

Next.js apps, route handlers, storage, auth-aware pages, and deployment.

06

Process optimization

Map manual workflows, reduce friction, and turn repeated work into systems.

// 04Systems // featured

Production systems and assistant workflows.

Real systems running today. Each entry maps a concrete problem to what was built and how it now operates.

SYS · 01AI Assistant

Hermes LINE AI Assistant

2026

Production AI assistant connected to LINE and deployed on a hardened VPS.

Problem
The assistant needed to run on real messaging traffic with secure public routing, service monitoring, and safe fallback behavior.
Built
A VPS-hosted Hermes workflow with secure routing, private service management, deployment health checks, and configurable model routing.
Result
A production-ready assistant foundation that can receive messaging traffic, report service health, and keep sensitive internals private.
HermesLINE Messaging APISecure Routing
Open dossier
SYS · 02DevOps Automation

Codex DevOps Companion

2026

Local Codex-to-Hermes workflow for alerts, reports, event logs, and production checks.

Problem
Operational checks, deployment notes, and LINE notifications were scattered across manual commands.
Built
A Codex-to-Hermes command workflow for health checks, notify, reports, event logs, and secured internal API calls.
Result
Repeatable local operations that let Codex report system state and send alerts without exposing private server internals.
Codex CLIHermes APISecure API
Open dossier
SYS · 03Knowledge System

Obsidian Vault Knowledge Assistant

2026

Read-only personal knowledge assistant built from a sanitized Obsidian vault export.

Problem
The private Obsidian vault should not be exposed directly to a server-side assistant.
Built
A sanitized, read-only knowledge export pattern for Hermes with personal assistant and wiki-style retrieval use cases.
Result
A safer assistant context path that keeps raw private notes local while still enabling structured knowledge lookup.
ObsidianHermesKnowledge Base
Open dossier
SYS · 04Full-stack App

CS Logbook System

2024

Full-stack logbook system for student activity tracking and admin workflows.

Problem
Student activity tracking and admin follow-up needed a structured workflow instead of scattered records.
Built
A full-stack logbook system with database-backed records, authentication workflows, dashboard views, and deployment operations.
Result
A maintainable student logbook foundation that supported administration workflows and later production handoff planning.
ReactNode.jsMongoDB
Open dossier
SYS · 05Landing Page

FlowSync SaaS Landing Page

2026

A standalone SaaS landing page prototype for an AI workflow product.

Problem
The portfolio needed a public-safe demo that shows landing page design, product messaging, and route-level layout control without exposing private systems.
Built
A standalone SaaS page outside the portfolio layout, with its own navbar and complete conversion-focused sections.
Result
A reusable demo that shows product-page execution, responsive UI composition, and clear separation between portfolio navigation and campaign pages.
Next.jsTailwind CSSTypeScript
Open dossier
// 05Process // build loop

How I take an idea to a system on a server.

A repeatable loop. Define the operator job, build the system, operate it in production, then iterate from real signal.

  1. 01
    Define

    Map the operator job

    Identify the recurring task, the failure modes, the trust boundary, and the success signal. No code until the loop is clear.

  2. 02
    Build

    Compose the system

    Wire AI, APIs, messaging, and storage into a workflow with strict secrets handling and a clean public surface.

  3. 03
    Operate

    Run it in production

    Deploy to real infrastructure with health checks, alerts, structured event logs, and a documented runbook.

  4. 04
    Iterate

    Tighten the loop

    Observe usage, prune friction, reduce surface area, and graduate one-off scripts into reusable system blocks.

// 06Stack // matrix

A pragmatic stack for shippable AI systems.

Tools chosen because they survive contact with production — not because they are trending.

AI & Models

//01
  • ChatGPT01
  • OpenAI API02
  • Claude03
  • OpenRouter04
  • Prompt patterns05
  • RAG-style retrieval06

Web & App

//02
  • Next.js01
  • React02
  • TypeScript03
  • Tailwind CSS04
  • Vercel05
  • Vercel Blob06

Ops & Channels

//03
  • LINE Messaging API01
  • PowerShell02
  • Codex CLI03
  • VPS · reverse proxy04
  • Health checks05
  • Docker06
// 07Manifesto // engineering

Operating principles.

How I think when I build — applied to every system I put my name on.

/01

Ship to production

A system that does not run in production has not been built yet. Demos are scaffolding, not the deliverable.

/02

Secure by default

Secrets live in env, not in code. Public surfaces stay public-safe. Private internals stay private — always.

/03

Automate the repeatable

If a task happens twice, it is a candidate for a workflow. If it happens three times, it already is one.

/04

Build for the next operator

Health checks, event logs, runbooks. The system should be operable by someone who is not me.

// 08Trace // experience

A working trace of systems built.

Three eras, all converging on the same theme: turn manual work into observable, shippable systems.

01

2026 - Present

Independent production systems

AI Automation & Systems Builder

Built and operated AI assistant workflows that connect LINE, internal APIs, VPS infrastructure, and personal knowledge systems.

  • Deployed Hermes LINE AI Assistant with secure routing, service monitoring, and health checks
  • Built Codex command wrappers for production health checks, LINE notifications, reports, and event logs
  • Prepared a sanitized Obsidian vault export for read-only assistant context
02

2024 - 2025

Project-based work

Full-stack Web Developer

Built web applications and dashboards using React, Node.js, databases, and responsive UI patterns.

  • Built CS Logbook workflows for student activity tracking and administration
  • Implemented authentication, database-backed features, and dashboard views
  • Delivered responsive interfaces with React and Tailwind CSS
03

2023 - 2024

Portfolio and product experiments

Frontend Developer

Created web interfaces, landing pages, and portfolio systems while building a foundation in modern frontend tooling.

  • Built React and Next.js interfaces with component-based structure
  • Practiced SEO-ready pages, responsive layouts, and content-driven sections
  • Used Git/GitHub workflows for project iteration
// 09Lab // showcase

Notes, demos, and works in progress.

Public-safe artifacts from the lab — explainers, demo surfaces, and exploratory builds.

Lab notes

Demos

// 10Channel // open

Build a secure, useful AI system.

I take on focused engagements where an AI workflow needs to actually run in production. Email is the fastest channel.

Start a transmission

chinnakrit.srp@gmail.com

Typical response within 24 hours

Secondary channels

All systems described stay within public-safe boundaries. No private infra, credentials, or operational details are exposed.

© 2026 · Chinnakrit Sripan