Tarun Kumar

THIS GUY BUILT A BOT THAT SEES LIVE SPORTS DATA ~8 SECONDS BEFORE POLYMARKET UPDATES AND TRADES THE LAG.

What is an interesting fact about React.js?

China just put a household robot in real homes.

IF you have less than 1k followers, Say hi, let's follow you immediately #follow

def create_multipliers(): return [lambda x: x * i for i in range(5)] multipliers[0](10) > 40 (not 0) All lambdas use final i=4 This is late binding in closures. Fix: default arg 'i=i' Ever hit closure bugs in list comprehensions? #Python #Coding #AdvancedCoding #Bug

JavaScript being JavaScript: "0" == 0 → true false == 0 → true "" == 0 → true But "0" === 0 → false Never use ==. Always === What's the weirdest coercion you've seen? #JavaScript #Coding #WebDev #Bug

Why do programmers prefer dark mode? Because light attracts bugs. 🐛 😂 #Coding #Programming #DevHumor #100DaysOfCode

Nvidia drops Nemotron 3 Super - a new open model optimized for complex agentic AI systems. Key highlights: - Hybrid Mamba + Transformer architecture with Mixture-of-Experts - Massive 1-million-token context window - Faster inference and better efficiency for long-running tasks - Released with open weights + datasets Designed specifically for multi-agent systems that need strong reasoning, coding & long-context handling. Nvidia keeps pushing the infrastructure edge for agents. Who’s excited to test this? #Nvidia #Nemotron #AgenticAI

Google upgrades Gemini for Workspace - now pulls data from emails, chats, files & web to create full documents, spreadsheets & presentations. - Turns Google Drive into an active knowledge base - Natural language prompts > fully formatted drafts - Uses multiple specialised models (including Gemini 3) Productivity just got a big boost for Workspace users. Repetitive tasks like financial analysis or presentation prep are getting automated fast. This is Google making AI deeply embedded in daily work. Using Gemini in Workspace yet? #GoogleGemini #GoogleWorkspace #AIProductivity

Microsoft is now prioritizing adaptable "generalists" over narrow specialists in many roles. - AI handles more routine/coding tasks > Companies want people who can work across domains and oversee AI tools Meanwhile, demand for AI Engineers remains one of the fastest-growing job titles globally. But layoffs continue in some areas as AI automates parts of workflows (seen in recent Block & Atlassian moves). Upskilling in agentic AI, prompt engineering & oversight skills looks key this year. Are you shifting your career focus toward AI? #AICareer #TechJobs2026 #FutureOfWork

Big chatter right now around Claude Code: - Anthropic added “Auto Mode” - AI can autonomously handle safe operations while pausing on risky ones. A balanced middle-ground between full autonomy and strict permissions. - Leaks surfaced about an unreleased model internally called **“Claude Mythos”** (capability assessments and dev details exposed via public cache). Lots of discussion on safety vs. ops security. - Ongoing March promotion: Doubled usage limits during off-peak hours (until March 28) for all plans. Claude Code and Cowork features (computer use, interactive apps) keep getting stronger for developers. What do you think about the new Auto Mode in Claude Code? Useful or still too cautious? #ClaudeAI #Anthropic #ClaudeCode #AICoding

Next.js 16.2.1 canary builds continue rolling out with bug fixes and performance tweaks. Recent activity includes: - Turbopack improvements (faster dev server startup, better HMR recovery after sleep) - Fixes for adapter outputs, metadata routes, and layout segment optimisation - Ongoing AI-related enhancements from the March 18 “Next.js 16.2: AI Improvements” release Next.js 16.2 brought faster rendering (25-60% in some cases), redesigned error pages, and deeper AI tooling support. Security note: Make sure you're on patched versions (16.2.x or latest 15.x) due to earlier React Server Component issues. Anyone upgrading to the latest Canary today? #NextJS #NextJS16 #WebDev #React

Final scoreboard: • Claude: 4/5 • ChatGPT: Most consistent • Gemini: Best for real-time research

Task 5: Real-time info & web search • ChatGPT: Good (especially with browsing) • Claude: Limited real-time access • Gemini: Best for fresh, up-to-date info 🏆 Winner: Gemini

I tested ChatGPT, Claude, and Gemini on the SAME tasks. The winner surprised me 🧵 Task 1: Write a cold email • ChatGPT: Felt generic, like every inbox email • Claude: Natural, human, actually personal • Gemini: Decent but too formal 🏆 Winner: Claude

"Vibe coding" is going mainstream. Qwen 3.5, Cursor Composer 2, and Google AI Studio are letting devs generate full-stack apps in natural language. The debate is no longer "should I learn to code" - it's "should I learn to prompt." 🤔 #VibeCoding #AIDevTools #CursorAI #LearnToCode #SoftwareDevelopment

Alibaba's Qwen 3.5 9B model runs on your LAPTOP, scores 81.7 on GPQA Diamond, and beats OpenAI's 120B model. The "you need a supercomputer for AI" era is officially over. Local AI coding assistants just levelled up. 💻 #VibeCoding #Qwen35 #LocalAI #AITools #Coding