Challenging Tasks and How Team Members Found a Solution

Challenges might not always be fun, but it’s very satisfying when one succeeds in overcoming them.

Whether one is appreciating the view from a mountaintop or admiring one’s newly-painted living room walls, the hard work required to get to that moment probably plays into how much fulfillment one feels.

We asked two of our Swan team members to share a challenge they faced at work and how they overcame it.

Yevhenii: The Challenge Solved with the Right Tool

I have a project in which there was a rather difficult task, although it was not a too cumbersome and large scaling type of task. The problem was with the use of AI technologies in it.

If I describe it briefly, in the client’s business logic, there was an intermediate stage that required human participation, which slowed down the process a lot. For example, the entire execution of one cycle of the business process would take about 10 seconds or something like that. But before that, the human being must complete a number of tests that’s related to the process of analyzing and splitting one large document into several small ones based on its content for subsequent processing.

So, it was decided to assign these responsibilities to Artificial Intelligence in order to fully automate the process.

As it’s turned out, the AI can easily analyze the pages in the form of images. AI could easily draw a conclusion about what is indicated on it, why this document is needed, what the proposal did, and the like.

But when it came to the moment of dividing the document into specific pages, the AI’s confusion began.

For example, if we had the document that consists of 40 pages, the eye could easily keep track of its numbers with the page numbers of 50, 70, and beyond of that, which certainly could not be drawn. And even if you ask the AI ‘why did you specify the page numbers that go beyond the boundaries of the original document’, the usual answer would be, ‘well, sorry, you’re right, here is the correct answer.” Without any explanation of the logic behind that.

The problem was not in compiling the product request, as it’s turned out, but in simultaneously processing a huge amount of pages. Because of which, the AI began to retrain itself with new, non-existent rules.

The solution turned out to be effective. It is now to split one large document into several sub-documents that uses not one request to the AI, but several. So instead of having 40 pages, we would have 20. This process is called decomposition.

And it’s even an excellent key to solve many problems — not only in programming but also in ordinary life problems. Because, you know, the pie is not eaten as a whole, but in pieces.

Maksym: The Challenge Solved by Teamwork

One of the most challenging task for me was work estimation.

Previously, I’ve worked as the only QA engineer on a project, and I had to deal with both manual and automation testing, and task estimation was not a part of my job. But now, it is one.

I have to estimate my tasks, and I try to do it as precisely as possible. The challenge here is that my current position requires more programming skills and framework knowledge, and some tasks may be not fully clear for me. Or I might have some questions about particular requirements.

As a result, tasks may be underrated or vice versa.

How do I deal with this challenge now?

As there are some other QA engineers in my team with more experience, I have a great example of how other guys solve similar problems. And also, I can ask my questions at any time, and thanks to my colleagues, get a helpful answer.

Solutions

At Swan Software Solutions, we create custom solutions to help businesses succeed. If you’d like to discover more about how we can help with your technology needs, schedule a free assessment.