Deployment & observability.

Getting it live, and knowing the moment it stops working.

Start a project How I build it
  • Docker
  • AWS EC2
  • Langfuse
  • OpenTelemetry

The problem

An AI system can break without crashing. Answers get worse, costs creep up, a tool starts timing out, and the first sign is a user who stops coming back.

Here’s how I build it

How I build it.

5 steps from the first conversation to launch. The diagram fills in as each part gets built.

  1. Step 01

    Package it

    The app and everything it depends on go into a Docker container, so it runs the same on a laptop and on a server.

  2. Step 02

    Automate the release

    A CI/CD pipeline tests and ships every change, so a release doesn’t depend on one person’s machine.

  3. Step 03

    Deploy to the cloud

    It runs on AWS or Azure, with secrets and access set up properly and room to grow with your traffic.

  4. Step 04

    Trace every call

    Each LLM call is traced: the prompt, the tools it used, how long it took and how many tokens it cost. Langfuse and OpenTelemetry collect it.

  5. Step 05

    Watch and alert

    Dashboards show the numbers that matter, and alerts fire when errors, response times or costs move outside their normal range.

What you get.

All of it, or only the part you need. Each piece can be scoped on its own.

Built with

  • Docker
  • AWS EC2
  • Langfuse
  • OpenTelemetry
  1. 01Containerisation and CI/CD pipelines
  2. 02Deployment to AWS or Azure
  3. 03Tracing and token/cost tracking on every LLM call
  4. 04Alerting and dashboards on the metrics that matter

Where I’ve done it.

Quoted from my CV, then the builds you can open and check for yourself.

  • Padelos

    Associate AI Engineer

    Jan 2026 — Present

    • Migrated the platform chatbot from Azure to AWS EC2 and shipped services with Docker and Bitbucket Pipelines CI/CD, adding CloudWatch-based observability.

Builds.

All projects

Questions clients ask.

Not covered here? Ask me directly.

Can you move our app to another cloud?

Yes. At Padelos I migrated the platform chatbot from Azure to AWS EC2, with Docker and CI/CD around it.

AWS or Azure?

Whichever you already use. If you use neither, the choice comes down to your other tools and the cost.

What does LLM observability show me?

Every call to the model: what was asked, which tools ran, how long it took and what it cost. Over time it shows whether quality, speed and cost are going the right way.

Can you add CI/CD to an existing project?

Yes. Tests and an automated release can be added to a codebase that already exists, without rewriting it.

Other services.

Work with me

Have a project like this?

Tell me what you want to build, and I'll tell you how I would approach it.

Let’s talk