The leak showed how quickly code can be copied once AI turns appropriation into ordinary workflow. A proprietary codebase became public infrastructure before breakfast At 4 AM on March 31, 2026, a developer in Seoul woke up to a phone full of notifications. Claude Code’s source had leaked. Within hours the code was mirrored across GitHub, screenshotted across social feeds, mocked as “vibe-coded garbage,” and ported into new repositories by people who had not worked at Anthropic the day before. By the time many engineers on the U.S. East Coast opened Slack, the incident had already passed through several distinct phases: exposure, ridicule, cloning, and legal containment. ...

Good NLP starts with reading the text: define the task, keep meaning-bearing words, then build simple, inspectable baselines. Some Text Problems Punish Shallow Reading Suppose you need to flag notes that describe active chest pain right now. The task sounds easy until the notes arrive. Reports chest pain x2d, worse on exertion. Denies chest pain; shortness of breath improved. Family history of CAD. No active chest pain today. Chest pain improved after nitroglycerin. Cannot rule out infection. No known drug allergies. SOB with exertion. Family hx diabetes. These fragments share vocabulary, but not meaning. Some describe active symptoms. Some are negated, historical, or about family history. Some express uncertainty. Some hide ordinary clinical meaning inside abbreviations. A model that treats them as ordinary word overlap will make confident mistakes fast. ...

Spec-driven development is older than AI, but agent workflows make the boundary between design artifact and executable logic newly important. (Source: “A very comprehensive and precise spec” from CommitStrip) An Older Engineering Habit Under New Pressure Picture a platform team reviewing an internal agent workflow for customer support. They are debating a scheduler that fans out work across multiple agents, retries failed steps, persists thread state, and hands off certain actions to a ticketing extension. One engineer wants to write the logic directly. Another wants a design document first. A third wants an RFC with interface contracts and failure semantics before anyone touches the service loop. ...

While the headlines scream of mass layoffs, the data suggests a more subtle and structural shift: the ladder is being pulled up from the bottom. Source: https://www.anthropic.com/research/labor-market-impacts (Figure 2) The Ghost in the Unemployment Statistics “AI is coming for your job.” It’s the refrain of every tech influencer and doom-scrolling headline. Two days ago, a report dropped that seemed to confirm the worst. Anthropic, one of the leading labs behind the very tech in question, released Labor Market Impacts of AI: A New Measure and Early Evidence. ...

We disentangle population risk from model-specific performance when subgroups enter with different risk profile. (Core idea of the adjusted TPR; some image components generated by NanoBanana 2) In part 1, we saw that a raw true-positive-rate gap can be a mirror of baseline risk differences, not just model behavior. In part 2, we make that distinction operational and show how to measure fairness once those risk differences are held constant. ...

We dive into whether and how forcing equal outcomes in healthcare algorithms can overlook the reality of baseline risk and demographic diversity. Clipart generated by Google NanoBanana 2 A Familiar Evaluation Result — and an Uncomfortable Question Let’s begin with a short vignette. A healthcare analytics team is reviewing a mortality risk prediction model used to trigger palliative care consultations. The model has been deployed for several months. As part of routine governance, the team evaluates fairness across demographic groups. ...

A language-first view of modern AI systems (2026) (source: Companies Bring AI Agents to Healthcare (WSJ)) A system that sounds simple—until it isn’t Imagine a health system designing a conversational AI service for telemedicine. Patients describe symptoms, concerns, and fragments of medical history in free text. The system responds conversationally, drawing on prior encounters, internal documentation, and clinical guidelines. It answers routine questions, summarizes relevant context, and—when appropriate—routes cases to clinicians. ...

Opening Happy new year! In my previous post, My Winter NLP Journey, I wrote about motivation: why I wanted to build a old model like GPT-2 and why implementation feels like the fastest path to understanding. I am glad to report I completed the journey as planned. This post is a reflection on what actually changed in my head after working through the course on my own. The biggest gain was simple to name but hard to achieve: several ideas that were vague for years finally became clear. ...

The Gradient That Changed Everything It’s almost coincidental. On the Christmas Eve this year, I came across a math problem asking me to compute the partial derivatives of Word2Vec’s naive softmax loss — standard fare for any NLP course. But something compelled me to keep going, to really understand what these update rules were doing. The result was deceptively simple: $$ \frac{\partial J}{\partial v_c} = -u_o + \sum_{w\in V} \Pr[w|c] \, u_w = U(\hat{y} - y) $$What struck me wasn’t the math itself albeit it’s elegant but straightforward. What caught my attention was the structure of the learning process. What this math formulation suggests is that updating the center word vector $v_c$ requires knowing the current state of all context vectors $U$. But updating $U$ requires knowing $v_c$ (illustrated by partial derivative regarding $U$–the other piece of the puzzle). This chicken-and-egg dependency — where each parameter set treats the other as temporarily fixed — reminds us of Expectation-Maximization algorithms. It isn’t EM in the formal sense, but the alternating dependence—treating one parameter block as fixed while updating the other—shares the same structural intuition. ...

An FP Practitioner’s Perspective on the Cloudflare Feature-File Outage Disclaimer: I am not affiliated with Cloudflare. This analysis is based entirely on publicly available incident reports and technical discussions. The architectural reconstructions and code examples represent my educated interpretation of what likely occurred, informed by functional programming principles and my own experience with similar failure modes. Where I speculate beyond published details, I have tried to make those inferences explicit. ...