Early in your career, it can sometimes feel like you’re not doing truly original work. With coding agents in our workflows, much of what we do looks like re-implementing solutions that already exist.

But every now and then, you face a problem where no ready-made solution exists and even AI can’t help you. And that’s when you realize you are solving real problems.

I remember one such case. I had to repeatedly append to an array of objects in a very large JSON file without reading the entire file into memory. Reading it fully would have added a huge performance cost.

The challenge was that Python’s append mode only allows writing at the end of a file, which would break the JSON schema. JSONL could have been a workaround, but the business constraint was to stick with plain JSON.

After some deeper thought, I realized I could open the file, seek to the last character, and inject the new object there. It may sound like a small challenge, but solving it felt original.

At the end of the day, that’s what keeps us motivated: the rare moments where you stop, think differently, and solve a problem that doesn’t have a straightforward answer.