ADAS Dev @Ford 🚙🛠️ Data Pipelines and MLOps by 🌞 Dancing my a** off and tweeting nerdy shit & @ManUtd by 🌚🇯🇲🇹🇹 Prev: @Veoneer Ambassador & Alum @UDMces
There is no "right" way to become a developer.
I worried how I'd apply web2 learning from robotics, image-processing, and data tasks - to the web3 world of smart contracts 🦀, UI//UX, and infra.
The more I learn, the more I realize there's space for devs of EVERY kind
🧵/
Que Sera, Sera.
Pour into things that pour into you. There’s no medal for curating a space you don’t get to use. No award for learning and not growing.
In the end it’s you and the shit you dealt with - the person you’re most accountable to is yourself.
USA. A Mexican restaurant. We had not yet ordered anything, and the food was already arriving.
Chips. Salsa. Unrequested. Free.
I stopped the waiter. "We have not earned these."
"They just come with the table, man."
They come with the TABLE. In my land, hospitality is a debt. Every gift creates an obligation, weighed carefully, returned in the proper season with interest of feeling. Here, the gift arrives before you have even proven you can pay for dinner.
This is not an appetizer. This is a declaration: we trust you. Eat.
I ate with the gravity the moment deserved. And then — I must report this calmly — the basket emptied, and a new one appeared.
"Did we…?"
"Refill," the waiter said. "It's bottomless."
Bottomless. They have wells of salsa. The supply lines of this nation are beyond anything my ancestors imagined.
My friend warned me. "Don't fill up on chips, dude."
Too late. I had accepted three baskets. Honor demanded each one be finished — an unfinished gift is an insult. By the time my actual food arrived, I was a ruined man.
I was not hungry. I was not comfortable. I had been defeated by a courtesy.
Generosity that arrives before the request cannot be repaid. It can only be survived.
I know the rule now. I have made my peace with the basket. One basket. Two at the most.
Who am I deceiving. There is no number of baskets I would refuse. The trust of a nation is in that salsa, and I intend to honor all of it.
Why is everyone talking about the DGX Spark for unified memory but not the ASUS Ascent GX10?
Now that I’m expanding my platform to robotics projects the context/token limits on the 3090 across multiple codebases is starting to hurt 🥲
Public mirror on github btw (active dev)
Bournemouth Co-CEO at the Tampa fan fest having a chat about Iraola leaving???
I actually rate this tbf proper leadership not sending out someone else to answer the tough questions.
Having done a ton of experimenting with Qwen 3/3.5/3.6, Qwopus distilled 3.5, Nemotron 2/3, etc. on a 3090 with an ACTUAL codebase.
GPU maxis are NOT wrong about capability - but I’m not seeing enough talk about memory/context/token management or skill/tool selectors. BIG gap.
software engineering is absolutely not being automated right now.
I barely ever code any more but I have never been busier. now that coding is 80% automated, the limiting factor is my ability to design, comprehend, and safely change systems. it's insanely exciting
Ok, I think my experiment leaving AI working on stuff 24/7 ends here. It doesn't work. Code explodes in complexity, results are not that great, the AI can't get past hard walls (it is still completely unable to even *grasp* SupGen), and it is insanely expensive (spent ~1k over the last 2 days). The best results are on the JS compiler, mostly because it is familiar (compared to inets), but not worth losing control over the codebase.
I think the dream of having AI's working on the background and making real progress on things that matter (i.e., truly new things) isn't here yet. It is still a machine hard-stuck on its own training data, incapable of thinking out of the box. It is great for building things that were already built. But not new things
Also coding normally has the under-appreciated advantage that you're doing two things at the same time: building a codebase *and* learning it. AI's do only half of that. The other half is obviously impossible 🤔
“A masterplan Jared’s been working on for 2 years” (Oct 25)
Kushner’s face when he revealed they planned rebuilding Gaza BEFORE Oct 7.
This was always the plan - it was never about Hamas
US special envoy Steve Witkoff & Trump son-in-law Jared Kushner.
The YouTube algorithm is recommending me videos about how cooked the cs new grad market is because I keep searching about how impossible home ownership is 😂😭
I knew it was bad out here but WOW affordability + new grad job market is COOKED.
This jobs report bro is holding…🥶
My phone finally force updated to IOS 26 and I have not considered going back to android this much since 2016. Genuinely the worst UI I’ve ever used AND it tanked performance/battery.
Get me out…
Gah damn it’s crazy what 6 days + (some) domain knowledge + @cursor_ai (Claude4.5 in @Copilot) + probably not a ton of sleep either 😅 can do. TLDR ⬇️
– Spun up self-hosted @MLflow + benchmarking
– Fine-tuned YOLO at 1280px (eval against popular models informed me multi-scale was the move)
– Realized 100+ epoch training on 5070 Ti was doable - 16gb VRAM a bottleneck doing ANYTHING during training w/o OOM
– Migrated MLflow + Ray + @traefik + Auth/@grafana /Ops stack to an old 2070 box
– Grabbed a 3090 Ti Founders on BF + reconfigured everything for dual-GPU
– Set up @Tailscale so I can monitor/trigger training on vacation 😄
AI accelerates a lot, but some domain knowledge still pays off — and yeah - tokens + hardware ain’t cheap — but it’s a pretty accessible way to learn if school isn’t your style. Will continue to update 🙏🏽
Focusing on PII Pro first as I find that more interesting atm.
Playing with a YOLO model and some WF/CCPD/RF datasets. Trying to see how far I can go with dev containers + WSL before I setup partitioning.
Dry training on 250k+ images surprisingly good before optimizations.