My first Competex challenge involved building a predictive model from a real-world dataset with incomplete and noisy data. I went into it thinking that selecting the best algorithm would be the biggest challenge, but I quickly realized that data preparation had a much greater impact on the final outcome.

I spent a significant amount of time cleaning the data, handling missing values, engineering meaningful features, and understanding the evaluation metric before making any modeling decisions. That process taught me that a well-prepared dataset often outperforms a more sophisticated model trained on poorly prepared data.

The challenge also reinforced the importance of experimentation and reproducibility. Keeping track of every change, validating results carefully, and documenting assumptions made it much easier to improve the model over time.

Looking back, the biggest lesson wasn’t about machine learning—it was about developing a structured, analytical approach to solving problems. That’s a mindset I’ve carried into every project since.