Fatih Cetinkaya

The weekend build: low cost and usable (sort of) local AI coding agent for people on a budget. Nvidia Tesla P40 (24GB), eGPU (Oculink) and a cheap mini PC (Intel N100). Running in VSCode via Ollama.

More Codex Usage Is All You Need

@mattpocockuk DeepSeek-Coder is not bad for asking questions. See the last image (VSCode) x.com/devfacet/statu…

Has anyone experimented with using local LLM's for codebase exploration? I probably wouldn't trust them to write the code (not good enough) but for exploring the codebase they'd do a great job at no cost and lower latency (depending on hardware).

Run the following command every 6–12 months on your unused SSD and NVMe disks or say goodbye to your data: sudo dd if=/dev/{device} of=/dev/null bs=1M status=progress iflag=direct

Is it just me or is this the worst Twitter/X “For you” tab algorithm update? It’s beyond bad.

Can we add a Text Mode to the @X app? It simply removes all photos and videos and only shows text in Dark Mode.

@karpathy Significantly less spam since I switched to T-Mobile + Aura

I get ~10 spam calls per day (various automated voicemails, "loan pre-approval" etc) and ~5 spam messages per day (usually phishing). - I have AT&T Active Armor, all of the above still slips through. - All of the above is always from new, unique numbers so blocking doesn't work. - I am on all Do Not Call lists. - I have iOS "Silence Unknown Callers" on, but even if it catches & silences them I still get the notifications. Not sure if other people are seeing something similar or figured out anything that works

AGI will not be built by anyone; it will bring itself into existence.

@solomonstre 🔥Lots of wisdom in that post. I bet lessons from Docker.

Apparently everyone wants to create their own AI protocol. My advice: focus on shipping instead. Solve a real problem for real users. Let the bureaucrats write the spec. If you grow fast enough, the specs will adapt to you. If you don't, no spec will save you.

VSCode is becoming an operating system for AI agents.

What if Grok could check clickbait links on X and mark them with a single tap?

@Jason @USPS I would pay USPS $100 a year not to deliver junk mail to my house.

If it was $10 a year to get @USPS to your home would you pay for it?

@elonmusk What if we could use a @grok prompt for the personalized X algorithm?

Please post feedback for improving the 𝕏 algorithm in replies. Be as specific as possible and include examples.

VSCode is to AI IDEs what Android is to smartphones.

I just cancelled my @Docker subscription due to the upcoming price changes. What are the alternatives for personal usage nowadays?

@TheZachMueller @bitdeep_ @cursor_ai check out the usage of '!' at #_pattern_format" target="_blank" rel="nofollow noopener">git-scm.com/docs/gitignore… You can ignore everything in the .cursorignore file first and then add the folders you want. You can also just use .gitignore file. They are both used for codebase indexing (without doing @) docs.cursor.com/context/ignore…

@devfacet @bitdeep_ @cursor_ai The question is how to do the add part (without doing @, I want passive not active lookup).

@TheZachMueller @bitdeep_ @cursor_ai why not ignore everything, then add only the folders you want?

@bitdeep_ @cursor_ai I want the opposite of an ignore though, that says what folders to exclude, no?

@aallan It's not his fault. It's the fault of the company that hired him to interview people.

That's a ridiculous thing to ask in this day and age. Genuinely would just walk out of the room.

@MitchHHL Cool! Do you still work on this project?

Snifferlyzer (Logic Analyzer + USB Sniffer based on RP2040) prototype :D

@elonmusk Why you don't create a kind of fund or grant where people in interest could apply and help with the Optimus mission?

Just did a walkthrough of the Tesla supercompute cluster at Giga Texas (aka Cortex). This will be ~100k H100/H200 with massive storage for video training of FSD & Optimus. Great work by the Tesla team!