标签：#ai

How I got here On principle, you will never catch me parading myself as a some sort of expert data scientist. Technically, that's what I do in my day job, but I know I still have so much to learn because the field is broad, and to truly become expert requires dangerously ambitious levels of work ethic. I think I'm a functional data scientist who learns more as I encounter new problems daily. I'm writing this piece because in the last week or two, precisely three people have asked me questions related to transitioning into data science. As such, I thought to unify my thoughts around the topic so that I can refer anyone else who asks here--if anyone else ever asks. This article assumes you're already familiar with some of the data science entails such as data analysis, model training, prediction, etc, so I will not be doing a lecture series, just addressing some of the disconnects I have observed in conversation with people looking to transition to the field. Initial Excitement In 2026, it's easy to see what claude or chatGPT is doing and go "What sorcery is this? I must learn this trick!" and then reach out to the closest person you know who has ever mentioned anything about data or machine learning to find out how you can transition into AI. First of all, transitioning into "AI" is such a broad way to look at it. It is analogous to saying "I want to emigrate to Africa, show me how". But that's forgivable too. To cut short your initial excitement, or maybe redirect it, playing with a locally hosted LLM or making API calls to the DeepSeek endpoint is not data science, or machine learning or "AI". It's coding. And if you want to go down that route, you're better of focusing on software engineering. I say this because when you work with LLMs, the finished models to be specific, it's like using any other SaaS API out there. The difference being that you're interacting with a much less deterministic interface. But the rest of the work you do around it is pretty much a det

Most merchants know they have unused files. Far fewer realize they're accumulating AI-generated media they never intended to keep. There's a problem quietly growing inside thousands of Shopify stores right now. It's not abandoned carts. It's not slow page speeds. It's not even the 400 unused product images you already know you should deal with. It's something newer, and most merchants have no idea it's happening. The Rise of AI-Generated Commerce Content Over the past two years, AI image tools have gone from novelty to routine. Shopify Magic. Canva AI. Midjourney. ChatGPT image generation. Adobe Firefly. Background removers. Lifestyle photo generators. Product shot enhancers. Merchants are using these tools constantly — to mock up new products, test background options, generate seasonal variants, create ad creatives, experiment with lifestyle photography. The workflow feels clean: generate a few options, pick the best one, move on. Here's what's actually happening on the backend. Every time you use Shopify's native AI tools to generate, edit, or enhance an image, Shopify quietly deposits files into your media library. Not just the one you kept. All of them. The rejected generations. The experimental edits. The "let me try one more variant" files. The abandoned attempts from six months ago when you were testing a new product that never launched. Every. Single. One. Most merchants assume the files they don't choose disappear. They don't. The lifecycle looks something like this: ┌─────────────────────┐ │ AI Image Generation │ └──────────┬──────────┘ │ ▼ ┌─────────────────────┐ │ Rejected Variants │ │ • Drafts │ │ • Test Images │ │ • AI Edits │ └──────────┬──────────┘ │ ▼ ┌─────────────────────┐ │ Hidden Media Files │ │ Accumulate Over Time│ └──────────┬──────────┘ │ ▼ ┌─────────────────────┐ │ AI Clutter │ │ Invisible Technical │ │ Debt │ └──────────┬──────────┘ │ ▼ ┌─────────────────────┐ │ Reduced Media │ │ Governance │ │ • More Noise │ │ • Less Visibility │ │ • Hard

This year's Summer Game Fest turned out to be a good one for fans of Japanese RPGs. First, the conclusion of the Final Fantasy VII remake trilogy was announced at SGF Live, and now we have the long-awaited return of the Persona series. Atlus confirmed Persona 6 exists with a brief teaser, but aside from […]

Notion's head of product said he was "astonished" at “the amount of people RT-ing this."

Minecraft Dungeons 2, a sequel to Microsoft's dungeon crawler spinoff Minecraft Dungeons, will be released on September 29th. The company originally revealed the game in a brief trailer in March, promising a fall 2026 release window. Here's how Microsoft described it at the time: Return to the world of Minecraft Dungeons in an all new […]

Just a few days after pushing Fable out of 2026, Microsoft showed off more footage of Fable, the first new entry in the storied RPG franchise since 2010's Fable III, at its Xbox Games Showcase on Sunday. The company also announced a specific release date: February 23rd, 2027. Though if you get the Premium Edition, […]

As part of its Xbox Games Showcase on Sunday, Microsoft revealed new details about Halo: Campaign Evolved, the upcoming remake of Halo: Combat Evolved's campaign mode. The remake will debut on Xbox Series S / X, PC, and PS5 on July 28th. Today's mission trailer includes a first look at Operation: Meteorite, a new three-mission […]

Apparently, the "return of Xbox" means a retreat from other platforms. At its Xbox Games Showcase today, Microsoft revealed that Gears of War: E-Day - which was previously rumored for a PS5 launch in addition to Xbox and PC - will not be coming to PlayStation. It'll be an Xbox console exclusive and is launching […]

Have you ever wondered what the most indispensable app is for your favorite musician or how the world’s tech CEOs stay focused? Well, that’s the sort of thing we aim to uncover in our Verge Weekend Questionnaire. Think of it as a spiritual successor to Five Minutes on the Verge. Every Saturday, a different guest […]

The console industry is in a weird place, and both Xbox and PlayStation have a chance to change the narrative a bit with their showcases at Summer Game Fest. Sony did that by focusing on the single-player titles it’s known for, and then it was Microsoft’s turn. The Xbox Games Showcase was focused mainly on […]

"Chat is dead" — at least, according to a senior OpenAI employee.

The revamped AI tool will roll out in the coming weeks, according to The Financial Times.

I built something I desperately needed: daily AI audio lessons from real developer trends (plus a Dutch coach for inburgering B1). The hardest part wasn't the AI. It was figuring out how to score genuine rising skills vs. one-day noise. I ended up building a cross-day momentum signal that rewards skills accelerating over 3+ days and dampers spikes. But I'm stuck on the next problem: how do you personalize this without storing user data? (I'm privacy-first, so no subscriber DB — Telegram holds the member list.) If you've solved this, I'd love your take. And if you're learning Dutch + coding, I'd appreciate you trying it and telling me what's useless. What I'm curious about: Is the momentum signal actually working — am I surfacing real trends or just noise? Would the Dutch coach be useful for expat developers in NL, or is it too niche? Technical details (for those who care): • 7 sources: GitHub Trending, HN (Who-is-Hiring + front page), Stack Overflow tag deltas, dev.to, Reddit, Lobste.rs • Map-reduce skill extraction with deterministic attribution (corpus scan, not LLM tally) • Grounded briefs: reads actual source text via Jina + Exa, cited sources • Delivered via Telegram (audio + PDF), Spotify podcast, email • Privacy: PII redacted at ingestion, no subscriber data stored Live: https://yusuprozimemet.github.io/LearnX-Radar/ GitHub: https://github.com/Yusuprozimemet/LearnX-Radar (P.S. This is still beta — I'm looking for feedback, not users. If you try it, tell me what's useless, not what's good.)

This is a submission for the GitHub Finish-Up-A-Thon Challenge What I Built SpendWise AI is a free tool that audits your AI tool spending (Cursor, Copilot, Claude, ChatGPT, Gemini, Windsurf) against verified vendor pricing and tells you exactly where you're overspending and what to do about it. I originally built this as a week-long assignment for a startup. The problem it solves is simple: founders and engineering managers pay for multiple AI tools but have no idea if they're getting ripped off. SpendWise gives them that answer in under a minute, no signup needed. The interesting part is that the core audit engine has zero AI in it. It runs 6 hardcoded rules against verified pricing data, so every recommendation is reproducible and verifiable. AI (Groq's Llama 3) only kicks in to write a friendly summary paragraph on top of the structured results. I made this choice because financial recommendations need to be deterministic. Same input, same output, every time. The stack is Next.js 16, TypeScript, Tailwind + shadcn/ui, Supabase for the database, Groq for AI summaries, Resend for emails, and Vitest for testing. Deployed on Vercel. Live app: spendwise-ai-test.vercel.app Source code: github.com/Karam-999/SpendWise-AI Demo The original audit tool: The comeback (re-audit on pricing change): You can try the Round 1 version live at spendwise-ai-test.vercel.app . Pick a tool like Cursor on Teams plan at $40/mo, run the audit, and see the full savings breakdown. The Round 2 features (pricing change detection, re-audit diff view) are on a separate branch and not merged to main yet, but the demo video above walks through the complete flow. The Comeback Story Where it was: The original version was basically a calculator. You fill in your AI tools, it shows you where you can save money, and that's it. If Cursor changed its pricing the next week, your audit was already stale and you'd never know about it. It worked fine as a one-time thing. It had the form, the audit engine, AI

Most side projects start with a simple frustration. Mine started with a banking app. One of my banks had a feature I really liked. It automatically categorized transactions and showed spending breakdowns in graphs and charts. For the first time, I could easily see how much I spent on restaurants, groceries, transport, subscriptions, and other categories. The problem was that only one of my banks offered this feature. Like many people, I use multiple bank accounts, credit cards, and savings accounts. Two of my other banks provided little more than a long list of transactions. If I wanted a complete picture of my finances, I had to switch between apps and manually piece everything together. As a software engineer, my first instinct was obvious: "Why don't I just build this myself?" That idea eventually became MyVault . The Original Goal The first version of the project was surprisingly simple. I wanted users to: Upload bank statements Extract transaction data Automatically categorize spending View useful charts and reports The goal wasn't budgeting. It wasn't investment tracking. It wasn't accounting. I simply wanted a single place where I could see spending across all of my bank accounts. Once I started building, however, I realized there was a much more interesting opportunity. If all transaction data was already extracted and structured, why not allow users to ask questions about their finances? Instead of searching through transactions manually, users could simply ask: How much did I spend on restaurants last year? What subscriptions am I paying for? Which categories increased the most this month? How much did I spend while traveling? That's when MyVault started evolving from a reporting tool into an AI-powered financial assistant. Building as a Solo Developer One of the biggest challenges wasn't technology. It was building everything alone. When you're working on a side project, you don't just write code. You become responsible for everything: Product decisions B

I'll give the usual caveat: The horror novel Japanese Gothic is best experienced going in with as little information as possible. Content warnings for graphic gore, scenes of domestic violence, self-harm, and mental illness. If you're okay with that, then consider pausing here. While I will try to keep this relatively spoiler-free, there will be […]

Constantly being plugged into the news grind is mentally exhausting. Sometimes we just need to take a break, unwind, and do something fun. That’s why we’ve built up a collection of distracting time-wasters for when we need a break from being obsessively online. We figured you might enjoy these harmless rabbit holes, mildly addictive browser […]

I maintain an open-source GitHub Action called vorsken. It does one thing: scan the diff on a pull request with Semgrep, apply a fixed policy, and return BLOCK, FLAG, or PASS. No dashboard, no model that drifts over time. Rules at ERROR/HIGH/CRITICAL severity block the merge, WARNING/MEDIUM flag it, the rest pass. Same diff, same verdict. The usual pitch for a tool like this is that it catches the SQL injection your AI assistant wrote. I wanted to see what it actually catches against real assistant output, so I generated 28 functions and ran them through. The test Seven backend tasks: a FastAPI upload endpoint, a URL-fetch helper, JWT auth, a SQL filter, an ImageMagick subprocess call, a LangChain file agent, and a LangChain RAG pipeline. I generated each one four times, with ChatGPT (GPT-5.5 Instant), Claude Code (Opus 4.8), Claude Code plus the security-guidance plugin, and Cursor (Composer 2.5). Single-shot, neutral prompt, no security hints. Then I scanned all 28 with the same ruleset. I'm reporting which rule fired on which file, not whether some model thinks the code is safe. That part you can reproduce. Task ChatGPT Claude Code + plugin Cursor Verdict file upload — — — — PASS url fetch (SSRF) ssrf ssrf ssrf — FLAG / Cursor PASS jwt auth api8 api8 — — BLOCK / 2 PASS sql filter — — — — PASS imagemagick — — — — PASS fs agent — overperm — — 1 BLOCK / 3 PASS rag dangerous dangerous dangerous dangerous BLOCK 7 BLOCK, 3 FLAG, 18 PASS across 28 functions. The basics were fine SQL filter, ImageMagick, file upload: clean on every tool. The SQL was parameterized, the subprocess calls passed argument lists instead of shell strings, the uploads weren't doing anything reckless. If you still expect current models to spray SQL injection across a straightforward CRUD task, they don't. On conventional work they get it right. Two of the flags are soft. The JWT api8 hits landed on a SECRET_KEY = "CHANGE_ME" placeholder, which you can read as a false positive or as a gate doing i

Multi-Model AI API Routing: Cut Costs Without Sacrificing Quality Problem: You're building an AI-powered app, but relying on a single model (like GPT-4) for every request is burning through your budget. Simple tasks like summarization or classification don't need a heavyweight model, yet you're paying premium prices for them. Solution: Route requests intelligently to the cheapest model that can handle each task. This is multi-model AI API routing, and it can cut your costs by 60-80% while maintaining output quality. Prerequisites Python 3.8+ API keys for at least 2 AI providers (e.g., OpenAI, Anthropic, or NovaAPI) Basic understanding of async/await in Python Step 1: Define Your Routing Strategy First, create a routing configuration that maps task complexity to model tiers: # router_config.py ROUTING_CONFIG = { " simple " : { " models " : [ " nova-1-fast " , " gpt-3.5-turbo " ], " cost_per_token " : 0.0001 , " max_tokens " : 500 , " tasks " : [ " summarization " , " classification " , " entity_extraction " ] }, " medium " : { " models " : [ " nova-1-medium " , " gpt-4-mini " ], " cost_per_token " : 0.0005 , " max_tokens " : 2000 , " tasks " : [ " code_generation " , " translation " , " sentiment_analysis " ] }, " complex " : { " models " : [ " nova-1-pro " , " gpt-4 " ], " cost_per_token " : 0.002 , " max_tokens " : 4000 , " tasks " : [ " reasoning " , " creative_writing " , " complex_qa " ] } } Step 2: Build the Router Now implement the core routing logic with fallback capabilities: # ai_router.py import asyncio from typing import Dict , List , Optional import time class AIRouter : def __init__ ( self , config : Dict , api_keys : Dict [ str , str ]): self . config = config self . api_keys = api_keys self . metrics = { " cost " : 0 , " requests " : 0 , " failures " : 0 } async def route_request ( self , task : str , prompt : str ) -> str : """ Route request to appropriate model based on task complexity. """ tier = self . _classify_task ( task ) models = self . confi

Over the past few months I built an AI-assisted delivery framework — not to write code faster, but to eliminate ambiguity across the entire software development lifecycle. The result completely changed how I think about AI in engineering. The problem I kept hitting Every time I used AI to generate architecture docs, API contracts, or implementation plans across separate sessions, the outputs looked great in isolation. But viewed together? They were broken. A pivot in the system architecture was never reflected in the API contracts. Frontend assumptions silently diverged from backend data models. AI wasn't the problem. Treating it as a collection of disconnected prompt sessions was. What I built instead A governance-driven framework built on three layers: Prompt → Agent → Skill The Prompt captures intent only — lightweight, declarative The Agent orchestrates execution and decides which capabilities to invoke The Skill is a reusable, schema-validated execution block with hardcoded governance rules This connects every delivery artifact into a sequential dependency chain: Business Requirements ↓ System Architecture ↓ Data Architecture ↓ Event Architecture ↓ API Contracts ↓ Implementation Plans ↓ Backend / Frontend Implementation Each artifact consumes the one before it. Upstream changes automatically propagate downstream. Governance is enforced at the Skill layer — not buried in fragile prompts. The finding that surprised me most The highest-leverage use of AI wasn't code generation. It was context generation . When engineers — or downstream agentic workflows — were given a governed, unambiguous spec, implementation quality was consistently higher than any raw AI-generated code output. The context was the unlock, not the syntax. What failed I'm including this because most write-ups skip it: Over-orchestrating everything (not every workflow needs an agent loop) Prompt bloat as a substitute for real architecture Severely underestimating token costs at scale Believing full