Cracking the code on complexity

Some problems don’t just feel hard. They feel impossible.

In my role as a Product Expert, I work with models that plan supply chains. On paper, it’s math. In reality, it’s millions of decision variables interacting in ways that aren’t always obvious. At that scale, complexity doesn’t just grow, it explodes!

That’s why, at OMP, solving these problems isn’t just about being clever. It’s about combining deep technical thinking with curiosity, collaboration, and the drive to make a real impact.

And recently, that mindset helped our team crack something that had been holding us back.

When things stop scaling

A lot of the problems we work on are NP-hard. In simple terms: they get out of hand, fast. At the extreme, even a model with just 25 binary decisions can mean exploring more than 33 million possibilities. That’s not something you can realistically solve when decisions need to happen quickly.

We had known about this possibility for a while, but the assumption was that customer supply chains were too connected and had to be solved together.

Still, one question kept coming back: did it really have to be this way?

The “hallelujah” moment

Then we started noticing something interesting. Not everything in these models is actually connected.

Some parts, what we call disconnected components, don’t influence each other at all. Yet we were still solving them together as if they did.

And that led to a simple but powerful idea: what if we didn’t?

A smarter way forward

We built an algorithm that automatically detects these independent parts and splits the problem into smaller subproblems.

No extra input, no model changes. It just works, and the impact was immediate. Problems that once took more than 24 hours can now be solved in seconds.

What it means to us

Being able to solve complex problems is our passion. Collaborating with and learning from each other adds tremendous value to the way we work and helps us tackle any challenge we face.

And there’s always another challenge waiting to be solved.  Curious how your skills could help?