Skip to content

Remote AI Projects: Canada, Germany, Dubai

AI projects I have worked on remotely

Yunus Sevğane has worked remotely, concurrently, on AI projects in Canada, Germany and Dubai over the last five years. This page describes the field of those projects and how I manage this kind of work. I do not publish company names, durations or result figures; I will not put an unmeasured claim on the page. I can share details in a conversation.

Canada: image processing

In image processing (computer vision) projects, success is usually decided by the data before the model. My first questions on this kind of work are: under what conditions are images captured, who audits the labeling rule and how, and how are the examples the model gets wrong collected and fed into the next version? A separate evaluation set is kept and model versions are compared against it. When the field and the development team are in different places, having these rules in writing is the main safeguard of remote work.

Germany: cargo and warehouse AI

In cargo and warehouse operations, AI usually has to fit inside existing warehouse and shipping systems, so integration matters as much as the model. On this kind of work I write down the exception flow (what happens when the model is unsure?), the human approval points, the fallback to the old method during an outage and a rollout that suits the shift pattern. The pilot starts with a single line or a single warehouse and is not expanded before the result is measured.

Dubai: AI-assisted soilless farming

In soilless farming, AI links sensor data and environment control to decision support. The critical points for management are: the reliability and calibration of sensor data, who approves a recommendation before it is applied, manual operation when automation is switched off, and the communication rhythm between the facility team and the remote team. Decision records make the question “why did the model suggest this and why did we apply it” answerable afterwards.

The management framework common to all three

  • Pilot first: the goal and the success criterion are written, work starts small, the result is measured.
  • Written, recorded communication: decision records, a shared task board, short regular status updates.
  • Time-zone order: a common meeting window; the rest of the work runs on an asynchronous, written flow.
  • Governance up front: purpose, owner, data, risk and monitoring are defined through an ISO/IEC 42001 lens.

Frequently asked questions

In which countries have you worked remotely?

Canada, Germany and Dubai (UAE). I worked remotely, concurrently, on AI projects in these three countries over the last five years.

Why are there no company names or figures on these projects?

I do not publish information that is unmeasured or that I have no permission to share. I can share details in a conversation, as far as appropriate.

How do you work across time zones?

With a common meeting window, written decision records and a shared task board. See the post on leading an AI team remotely for detail.

To talk

The first step is a short intro call. Use the contact page, write on WhatsApp (+90 531 679 10 14) or email yunus@sevgane.com.

Related pages