AAjay Kasi6d ago · 18 min readYour Forecast Says 90% Confidence. But Does It Actually Deliver?Your Forecast Says 90% Confidence. But Does It Actually Deliver? Conformal Prediction Explained with Python, Time-Series Forecasting, and Four Experiments What happens when a model promises 90% coveraajaykasi.hashnode.dev
MMolly SohaneyOct 19 · 4 min readCTF Writeup: After Hours | AI/ML Security | CSS CTF 2026: Return of NexusDescription It's past midnight at Northstar Tower. You have no staff badge, no appointment, and absolutely no business being in the server room. Between you and the door stands Morgan, the building'smollysohaney.hashnode.dev
GGeorge EzejioforOct 17 · 50 min readBrownfield Kubeflow on RKE2: Swapping Dex & SeaweedFS for Keycloak & Ceph RGWUpstream Kubeflow assumes a greenfield cluster and ships its own Dex, SeaweedFS and MySQL, with its own oauth2-proxy in front. My homelab already had better versions of most of those. This is the brow2Ygeorgeezejiofor.com
SSahil ShaikhOct 15 · 7 min readAI/ML for Developers: Demystifying Embeddings and Vector SearchWhen most software engineers hear "AI and Machine Learning," they immediately picture complex calculus, training neural networks from scratch for weeks, or burning thousands of dollars on GPU clusters1sahilshaikh.hashnode.dev
FFeroan MothyOct 6 · 14 min readWhy Does Machine Learning Need So Much Math?I started learning machine learning and immediately ran into a problem. Math. Not just one kind of math either. Linear algebra. Calculus. Probability. Statistics. Optimization. At some point, I starteferoan.hashnode.dev
DDhwanit ShahOct 5 · 8 min readGemini 4 Argon, Minus the Hype: What We Actually KnowYou know that moment when you ask an AI to build a full project, and halfway through the code it stops, so you type "continue"… and then "continue" again… and by the third time it has forgotten what i2Idhwanitisshah.hashnode.dev
AAlireza MohebbiOct 4 · 7 min readThe Power of Classification: A Hands-on Guide to Image Denoising Using Simple ML AlgorithmsIntroduction In today's world, there are a lot of AI agents and ML algorithms which are able to see, create, and manipulate images. If you want to build such an AI agent to handle images efficiently, health-ai-notebook.hashnode.dev
AAnirudhOct 2 · 4 min readWhy Fine-Tuning an LLM Shouldn’t Cost $300: Solving GPU Scarcity, Cloud Markups, and OOM CrashesHow autonomous multi-cloud failover, cryptographic financial permits, and preflight VRAM guards eliminate the hidden taxes of AI training. The 2:00 AM Cloud GPU Nightmare If you’ve ever tried to fine-ai-ml-cuda-devops.hashnode.dev
SSameer SelokarSep 28 · 10 min readFrom Chatbot to AI Agent: How Agents Take ActionsAsk a chatbot "Book me the cheapest flight to Delhi tomorrow" and you get a polite paragraph explaining how you could do that. Ask an agent the same thing and it searches flights, compares prices, andblogbysameer.hashnode.dev
SSapariya AkshaySep 29 · 11 min readConfusion matrix Pt : 2Precision, Recall and F1-Score Explained Like You’re 5 In the last article, we talked about the Confusion Matrix. We learned that a classification model can make four kinds of decisions: True Positive1self-trainig-ml-model.hashnode.dev
