Applied AI Engineer · Dubai

Applied AI engineer

An applied AI engineer builds language models into software that does a real job, with tests around it. I build retrieval with citations, agents that hand off to a person, evals and guardrails, and MCP servers.

What I build

Four kinds of system

RAG with citations

Answers from your own documents, each one showing its source, and a plain "I don't know" when the documents don't say.

See it in Open Brain

Agents with a human check

Agents that score their own confidence and hand the case to a person when it falls below the bar.

See it in Kredal

Checks that don't trust the model

The model reads, plain rules decide, and tests show when either one is wrong.

See it in Invoice Inbox

MCP servers

Tools that Claude and other AI clients can call, behind an OAuth 2.1 sign-in.

See it in Open Brain

Start here

Three projects to check

  1. Invoice InboxThe model reads, plain rules decide. 24 of 24 planted errors caught.
  2. Open BrainCited answers or a plain "I don't know", over MCP with OAuth 2.1.
  3. KredalEight Gemini agents that ask a person when confidence is low.

FAQ

Questions

What does an applied AI engineer do?

They build language models into software that does a real job: retrieval, agents, tools, and the checks around them. Most of the work is making the model reliable on real data, not training it.

How is an applied AI engineer different from a forward deployed engineer?

A forward deployed engineer does the same building inside a customer's team, on their data. That's the role I want next, and the forward deployed engineer page covers it.

Which projects should I look at first?

Invoice Inbox, Open Brain and Kredal. Each project page says what I measured and what the numbers don't show.

Last updated 14 Nov 2026