InPython in Plain EnglishbySilva.f.francis·Sep 25Sorting Your Suitcase by Vibe: Learning K-Means Through the Way We TravelFour ways of getting somewhere, one algorithm that figures out the grouping without ever being told what a “mode of transport” is.A response icon1A response icon1
InPython in Plain EnglishbySilva.f.francis·Sep 24Stop Fighting GridSpec: How subplot_mosaic() Lets You Draw Your Figure Instead of Coding ItA layout engine that reads ASCII art — and quietly makes one of Matplotlib’s most annoying corners actually pleasant to use.A response icon1A response icon1
InPython in Plain EnglishbySilva.f.francis·Sep 23The Quiet Genius of MaxPooling: Why CNNs Throw Away Most of Their Own Data on PurposeA layer that discards 75% of its input every time it runs — and why that’s exactly the point.A response icon1A response icon1
InPython in Plain EnglishbySilva.f.francis·Sep 23Teaching a Computer to Recognize a Face: A Field Guide to Face Recognition in PythonFrom Haar cascades to Siamese networks — what actually works, what’s outdated, and what nobody tells you about deploying this responsibly.
InOperations Research BitbySilva.f.francis·Sep 21Packing Your Data’s Suitcase: A Practical, Slightly Opinionated Guide to PCAHow a technique built on eigenvectors quietly runs half of modern data science — and how to actually use it without breaking your results.A response icon1A response icon1
Silva.f.francis·Sep 21Why the Crowd Beats the Genius: A Practical Guide to Ensemble Methods in PythonI used to think of ensemble methods as a slightly cheap trick — train a bunch of mediocre models, average their guesses, and watch the…
InOperations Research BitbySilva.f.francis·Sep 20Beyond BLEU: Understanding METEOR, the Translation Metric That Actually Rewards MeaningThe first time I compared BLEU and METEOR scores on the same batch of machine translations, I found a sentence BLEU had scored close to…
InPython in Plain EnglishbySilva.f.francis·Sep 20Teaching Machines What We Actually Want: A Practical Tour of RLHFSomewhere between “the model does what the reward function says” and “the model does what people actually want” sits one of the harder…
InPython in Plain EnglishbySilva.f.francis·Sep 18Django’s Authentication System, Properly Understood: From Login Forms to Custom BackendsThe first time I shipped a Django app to production, I genuinely believed I’d handled authentication correctly because django.contrib.auth…
InGoPenAIbySilva.f.francis·Sep 18The Monolith I Should Have Kept: A Python Engineer’s Honest Guide to Choosing an ArchitectureCoupling, scaling, deployments, data ownership, and the operational bill nobody reads until it arrives.