Describe a complex problem you faced and how you solved it

In my first year, I was assigned a small reporting feature that looked simple, but the numbers coming out were inconsistent with an existing dashboard. What made it complex for me was that the data passed through several existing jobs, and the documentation didn't match what the system was actually doing. I started by reproducing the mismatch with a few specific examples and tracing the data step by step until I found that one job was rounding timestamps differently from the others. I checked that theory with my mentor before changing anything, then I updated the logic, added a test for the failing case, and verified the report matched the dashboard in the environments where it had been wrong. After that, I wrote down the investigation steps for myself because I learned that when a bug touches multiple components, I work faster if I first isolate where the behavior diverges instead of editing code right away.

At my internship with a small e-commerce team, I helped with a checkout flow update that seemed straightforward, but we started getting complaints that some coupon codes worked on desktop and failed on mobile. I was asked to look into it with my supervisor, and the tricky part was that the problem only happened for a few specific combinations of browser, tax rules, and discount type. I recreated the issue using test orders, compared the request data between working and failing cases, and found that one screen was sending the coupon code in a format the backend didn’t expect when the order total had certain decimal values. I fixed the formatting on the client side, added a few tests for the edge cases we had missed, and then retested the full checkout path with my teammate before we released it. What I learned from that experience was to always check the actual data being passed between steps, because the issue is often not where it first appears.