Concept of Large Financial Models
·December 17, 2025
When one AI model decides to become a specialist instead of a generalist: training a single model to obsess over one ticker until it knows it better than its own reflection.
The Effie Labs Blog
Also writing at Machine Learning Pills
ML Library Problems
Problems (fun or hard) in data science and machine learning—implementing things from scratch (preprocessing, gradient descent), theoretical questions, and subtle but important differences in library functions. For example, the difference between pd.merge() and pd.join() isn't just syntax. These are things I've learned the hard way, often after debugging longer than I'd like to admit.
When Broadcasting Doesn't Broadcast
·December 11, 2025
Understanding why NumPy throws shape mismatch errors even when arrays look compatible. How broadcasting aligns dimensions from the right.
Views vs Copies
·December 10, 2025
Why modifying a slice changes your original array, and how to tell when NumPy creates views vs copies.
Integer Overflow
·December 9, 2025
How NumPy silently wraps around when integers overflow, and why dtype matters more than you think.