2021 in a nutshell

Tom and Jerry and Offenses under IPC A Thread -

Learnt about an app called Onramp Money today! Users can turn fiat currency to #Crypto Also allows users to pay via #UPI or #IMPS and buy crypto in return! Any users so far? Lets discuss.. #CryptoIndia #Web3 #Web3Community

What we fail to see is that there is more to #Web3 than just #NFTs & #Metaverse #Web3 represents the next phase of the #Internet It focuses largely on user control, #privacy & elimination of intermediaries The future is not #Web3 vs #AI but both Ad Age #Web3MarketingSummit

Happy #BitcoinPizzaDay 🍕

India's leading #crypto exchange CoinDCX : India Ka Crypto Coach launches #DeFi wallet @Okto_app The wallet allows users to swap tokens and earn from vetted services It has also introduced an #AI based multi-factor authentication to secure self-custodial DeFi wallets

Backed by Coinbase Ventures 🛡️ Bain Capital & Pantera the app looks promising and is a solid move in the ecosystem! Good one Sumit Gupta (CoinDCX) Further @nrjkhandelwal said that the team has filed for patent for this technology Must say bright days ahead for #DeFi #CryptoUpdate #DeFiWithOkto

Anon, what’s stopping you from living like this?

People waiting for better coding models don't realize that the quadratic time and space complexity of self-attention hasn't gone anywhere. If you want an effective 1M token context, you need 1,000,000,000,000 dot products to be computed for you for each of your requests for new

How do you handle token limits when MCP API calls return massive responses during multi-step reasoning? 🤔 Single-step calls are manageable, but when following chains of thought or doing multi-step analysis, those tokens add up fast! We've implemented smart optimization

I think server side rendering created more problems than it solved.

Lol berghain challenge getting intense. With upcoming folks berghain.challenges.listenlabs.ai Haven’t touched code in 2 days. still learning all on luck, variance and smart algorithms lets see what comes next.

Tested Modal's H100s for hyperoptimization today. No waiting for instances to spin up. Ran my workload, got results, shut it down. The serverless GPU thing actually works without the usual headaches. Pretty solid setup. modal.com #aiinfra

Its slow you know it everyone knows it. stop making it a big deal. x.com/svpino/status/…

Waiting for posts: Switched from Codex to Claude code again x.com/claudeai/statu…