Anna Kashkina was born in Tashkent, Uzbekistan, and grew up in Russia coding at a time when it was still seen as an odd thing for a young girl to be doing. Six years and a career in competitive programming later, she is finishing her degree at Amsterdam Tech and quietly solving a problem that costs tech companies millions.

Here is that problem. When a new developer joins a company, it takes them three to nine months to get comfortable with the codebase. That is a full quarter to nearly a year of expensive salary going into ramp-up. Every team leader knows it. Almost nobody has fixed it.

Anna’s final project, called CodeProp (running at CodeProbeApp.dev), takes a swing at that number.

Where Her Real Insight Came From

She did not start with an idea. She started with research: forums, interviews, hours of listening to developers describe the awkward social dance of joining a new codebase:

• The fear of asking obvious questions.
• The lack of structured guidance.
• The exhaustion of walls of documentation that never quite explain what you needed to know.

Her first prototype was pure engineering flex. Sophisticated algorithms producing beautiful explanations. She tested it on a real codebase and watched users get tired instead of getting smarter.

That was the pivot moment.

The Re-think

“People want to feel professional and have control over their knowledge,” Anna says. “Traditional tools that just provide answers don’t build that sense of mastery.”

So CodeProp stopped explaining. It started asking. The AI now guides new developers through real code with questions, forcing active learning instead of passive reading. It is closer to a great mentor than a great manual.

This is not the kind of project you build after 12 weeks of theory-heavy lectures. It is the kind of project you build when your programme forces you to ship real things that solve real problems.

Why This Matters for Anyone Considering a Tech Degree

Anna did not come to Amsterdam Tech to earn a certificate. She came because she saw where machine learning was heading and wanted an international environment to build in. What she is walking out with is a working product, a portfolio piece that speaks louder than any transcript, and a piece of published thinking (her earlier Medium article on training a neural network to beat human Atari scores).

That is the difference between studying software engineering and doing software engineering.

If Anna’s trajectory sounds like the shape you want, take a closer look at the BS in AI and Machine Learning Engineering or the Professional Masters in Data Science and Leadership. Both push you toward the sort of final project a hiring manager will actually click on.