Muhammad Here @devvMuhammad

I've started to build a 1v1 coding platform in Elixir and Next.js Let's see how it goes

Cursor Autocomplete goes crazy sometimes!

Why do domain managers always have such shit interfaces? Like how am I even supposed to use it 😭

Claude Code planning mode is so so better than Cursor. It feels like I'm talking to an engineer #claudeai

Wouldn't that be so cool if Supabase could add a "Copy as Markdown" in top right so we can just paste it for LLMs? Setting up Server-Side Auth initially is a little work and LLMs usually mess up with this a lot. @supabase @jonmeyers_io

I implemented a text over image search. I used - OpenAI Clip for Embedding Images - FAISS for indexing the vectors I used a COCO image dataset from Kaggle. I'm looking to add a way to use different indices for the same dataset to compare each.

How big of a difference is there in Opus 4.1 v Sonnet 4?

I implemented a text over image search. I built it using FAISS and OpenAI CLIP Model. - OpenAI Clip for Embedding Images - FAISS for indexing the vectors I used a COCO image dataset from kaggle. I'm looking to add a way to use different indices for the same dataset to compare…

I wanted to implement text search over images. Hence, I stumbled upon - FAISS (wanted to use it for so long) - CLIP for generating text + image embeddings Can't wait to build out the UI for this

There is nothing more peaceful than seeing your environments pop up when selecting the interpreter path #Python

Which one is used at large scale tech companies? Servers with a Load Balancer (Horizontal Scaling) OR Serverless

Convex is actually so fun to use. You don't have to manually revalidate caches to get the latest data. It is realtime and syncs up with the latest version itself. Looking forward to using it in more projects! Thank you @convex_dev

Added some simple tool calls with the help of Vercel AI SDK It has made creating chatbot interfaces a breeze, both on the frontend as well as the backend. I just love it!

Finally built something that uses RAG. It wasn't difficult. It was pretty fun building it with Langchain, Pinecone, Convex and Vercel AI SDK. I want to learn about LLMs. I guess it's time for a Deep Learning Arc. Enough Web Dev. It can continue alongside What do you think?

I've been building a RAG app pretty myself w/ Langchain & Vercel AI SDK. I don't know about best practices. Do you guys embed every message to the vector database? How do you maintain context? Do you summarise the whole conversation? The whole idea is to minimise the token…

I gave Opus 4 a shot and it shot back! Didn't expect it was this expensive 🤣

Cursor had its own CLI but I just want to try out Claude Code so bad. Let's see how it goes with cursor and claude code both slaves working for you at the same time This is the best time to go out there and build stuff you want. What a time to be alive! #Cursor

Finally found a desktop to interact with my postgres databases that doesn't look trash No more pgadmins or mySQL workbench Thank you @popsql for existing

quit my 4k$/month job, im posting every shit here, its my first post lets see how things go with twitter