I am looking for a cracked and based software engineer who can build with NextJS and Supabase. Maybe a bit of Python for data pipelines. Must be a US citizen, preferred to reside in the US. This is a co-founder type position that will require long conversations to make sure we

I’m thinking of pitching a hybrid SaaS model to a few clients (base features for everyone paid by monthly subscription, custom features paid by development). Does anyone have any advice to using this type of model?

OpenAI just announced they're killing off API access to GPT-4o in February 2026. Yeah, the fan-favorite multimodal beast that everyone's been hooked on. Developers get a measly three months to switch to something like GPT-5.1, which sounds great on paper but let's be

NVIDIA just dropped Nemotron-Elastic-12B, and it's a smart play in a world where AI training costs are getting ridiculous. The big idea? One model trained once that can spit out 6B, 9B, or 12B variants on demand, no need to retrain or store separate beasts. That's huge for devs

Ever feel like the AI arms race is pricing out everyone but the big players? Perplexity AI just dropped TransferEngine and pplx garden, open source tools that let you run trillion-parameter LLMs on your existing mixed GPU clusters. No need to shell out for a fleet of shiny

Nothing like a government document about cybersecurity where they don't know which way the slashes go:

This is awesome and works really well!

Update: after extensive research I decided to buy a Mac Studio with 96GB RAM and a ton of CPU and GPU cores. Setting it up this week and will share benchmarks.

Unpopular opinion: Your kids can survive without screens Monday-Friday Mine do. Here's the system: Screens = weekends only • Max 1 hour per day • Must earn it (clean room, respectful behavior) • Break the rules? Lose it that weekend "But my kids would lose their minds!"

LLM hallucinations are not a reason to abandon the technology. They are an obstacle that can be overcome by using the right model, context engineering, and prompting techniques.

I really enjoy working with some of my current clients and coworkers. Many of them are LLM-hesitant or straight up anti-LLM. As a software developer and data engineer this is not a defensible position in late 2025. I am building PoCs, functional software, and efficient data

Weird experience at a Toyota dealership today. They gave me a price for my trade in using the KBB instant price feature. The offer was $18K but I know the KBB trade in range is $18.7-$21.9k. So I pressed him on it and found myself arguing the difference between KBB trade in and

Running self-hosted LLMs in your enterprise? GPU costs can bleed you dry if you're not careful. ScaleOps just dropped a new AI infra product that claims to cut those costs by 50-70% for early adopters, all by automating workload scaling without touching your code. We're

xAI just dropped Grok 4.1 Fast, and on paper, it's a beast for devs: handles 128K context windows, cuts down on hallucinations, and runs agent tools like web search and code execution at a fraction of the cost of competitors. We're talking $0.59 per million input tokens which is

I see the play now. Make it seem like Trump is trying to overthrow future elections and take permanent control of the government with illegal orders. It’s all vague enough to rile up the base but give them room to slip out of actual accountability. It’s the old broad brush

Someone please explain why Fireman is not the best job of the first responders. Seems like a no brainer to me.

I’m working with a different kind of python today.

The most expensive video I’ve ever made: https://t.co/hlgAVNP37F Unfortunately I left the Azure SQL Managed Instance running for 1.5 months after filming. 😬 The bill was an extra $1,200!

Just read this piece comparing Gemini 3 to GPT 5.1, and it's got me thinking about where AI is heading. It's not just about who's got more flops; it's the real-world strengths that matter. Gemini 3 shines in handling multimodal data, like blending text and images seamlessly,

OpenAI just dropped GPT-5.1-Codex-Max, and it's already handling 24-hour coding marathons internally. As someone who's seen AI tools come and go, this one's got my attention for its long-horizon reasoning; think debugging and refactoring over extended sessions without losing

The past couple of days I started playing with Power BI Paginated Reports again. It has been awhile, but the experience is the same. Kind of mediocre, overly complicated, not ideal. So I built a paginated report builder. It is legit! This allows you to connect to a data model,