I am a data guy, not a developer. I am, however, an AI guy. With the help of a few blog guides, Lovable
lovable.dev
Lovable.dev, and VSCode copilot, I have managed to build my first Next.js analytics dashboard app and deploy to Vercel. If I can do it, so can you.

For anyone thinking of switching to an OLED monitor and worried about text fringing, don't. I recently upgraded to a 27" 1440p QD-OLED and read all the Reddit horror stories. In my experience thus far, fringing is a non-issue. Although I can't speak to WOLED.

Ironically, we did include a pair of Nvidia GPUs in the build, so at least that's a bit of money for Jensen to offset the $500+ billion drop in market cap this week.

Having just finished a build of a new ML workstation for a Data Scientist in my team, we'll be jumping straight on #DeepSeek R1. Thoughts incoming over the next few days.

Having watched a couple of the RTX 5090 reviews, my hot take is that running DLSS 4 performance with frame generation looks like a stuttering, artifact ridden hot mess. Like, what's the point in being able to run Cyberpunk 4k @ 300+ FPS when it looks garbage?

If you are using or thinking about using LLM thinking models for code projects, I have found much better results from Gemini 2.0 Flash Thinking than I have from OpenAI o1-mini. Give it a try in Vertex Studio, it's pretty ninja.

I do complain about working LLMs but honestly, it's been a game changer over the last couple of years. A great recent example is #NotebookLM from Google. Being able to digest dozens of lengthy articles and YouTube vids as a podcast is pretty wild.

I feel like 50% of my workday is now spent manipulating LLMs into doing what they are told, or debugging AI python applications. I thought these things were supposed to make us more productive.

Everyone is very animated about the Nvidia keynote at #CES2025. Personally, the debate around DLSS 4 and Frame Generation is less interesting than the fact that the 5090 is a return to a duel slot card, rather than being the size of an overfed dachshund.

Exciting news from @hf.co to start 2025 with the introduction of Smolagents. It will be interesting to see how the framework matures over the next six months. It might even wean me off LangGraph. huggingface.co/docs/smolage...