Not enough CTOs are talking about how they are practically putting AI codgen and agents into their existing large teams of SWEs: tools, training, and culture.
Here’s a few things we’re doing:
- ripping down everyone’s egos about what can and can’t be done by whom (ai or human)
- allocating budget to experiments
- getting legal, infosec, and ops simplified and aligned early
- focused slack channel to share wins, fails, and asks
- 1 motivated sponsor per tool
- use it or lose it access
- calling on peers for use cases and learning
- ai transparency (no one pretends they’re not using a tool to do work)
What’s working:
- ChatGPT in non tech functions
- v0 for prototyping
- cursor in pockets
- copilot in pockets
- chatPRD for engineers
What’s not
- 8 million rogue meeting note takers
- think we’re underusing perplexity
- no formal l&d
- only a few pushing novel thinking
- minimal automations & agent work
I'm also at https://t.co/pnO4QcF0Mt
The algo changes here are making it harder to stay on. It's frustrating how the tech and Infosec folks are all still posting though. Your posts are a lot less discoverable now. :/
A "close all tab groups" and "open saved tab group I'll type the name of next" keyboard shortcuts would be neat too. Is there an API for this I can wrap with an extension? cc @__apf__@laparisa
Dear @googlechrome, we'd like to search across history, bookmarks, and open tabs using the address bar. Bonus point if you wire the LLM history search to this unified search.
The at history and at tabs can stay but maybe add an all encompassing at local or at past?
Dear @googlechrome, we'd like to search across history, bookmarks, and open tabs using the address bar. Bonus point if you wire the LLM history search to this unified search.
The at history and at tabs can stay but maybe add an all encompassing at local or at past?
If you have experience automating deployments and running cloud services on top of GPUs and/or other kinds of bare metal. My organization is hiring in CA, WA, and OR.
https://t.co/qXyGemcGsW
#NVIDIABlackwell has arrived. @OpenAI , we’re excited to see what you can do with one of the first engineering builds of the #NVIDIADGX B200 system.
Learn more about DGX B200 here: https://t.co/6oGaZuPkSV
NotebookLM is quite powerful and worth playing with
https://t.co/EMHIjc15iU
It is a bit of a re-imagination of the UIUX of working with LLMs organized around a collection of sources you upload and then refer to with queries, seeing results alongside and with citations.
But the current most new/impressive feature (that is surprisingly hidden almost as an afterthought) is the ability to generate a 2-person podcast episode based on any content you upload. For example someone took my "bitcoin from scratch" post from a long time ago:
https://t.co/7ajZNZ0BGi
and converted it to podcast, quite impressive:
https://t.co/ZZn0LJgsnu
You can podcastify *anything*. I give it train_gpt2.c (C code that trains GPT-2):
https://t.co/gDrAqix4Iv
and made a podcast about that:
https://t.co/bgcwmQr5d7
I don't know if I'd exactly agree with the framing of the conversation and the emphasis or the descriptions of layernorm and matmul etc but there's hints of greatness here and in any case it's highly entertaining.
Imo LLM capability (IQ, but also memory (context length), multimodal, etc.) is getting way ahead of the UIUX of packaging it into products. Think Code Interpreter, Claude Artifacts, Cursor/Replit, NotebookLM, etc. I expect (and look forward to) a lot more and different paradigms of interaction than just chat.
That's what I think is ultimately so compelling about the 2-person podcast format as a UIUX exploration. It lifts two major "barriers to enjoyment" of LLMs. 1 Chat is hard. You don't know what to say or ask. In the 2-person podcast format, the question asking is also delegated to an AI so you get a lot more chill experience instead of being a synchronous constraint in the generating process. 2 Reading is hard and it's much easier to just lean back and listen.
Cloud, Platform, dev, or SRE/CRE managers. If you want to work with us at @nvidia building large scale ML and GenAI infrastructure from the bottom up for ourselves, our partners, and our customers. Here's your chance:
https://t.co/jyI4EVifHf
Link to IC req on my profile.
First job req is out: https://t.co/Xz5O2VAvTY
I'll fix some formatting errors later this week and the mgmt one is also about to leave the oven.
Feel free to apply directly. I'm also happy to answer any questions.
We are looking for platform eng and SRE folks in CA and WA (various levels both IC and Mgr). DM me if you are interested.
BMaaS, Kubernetes, Slurm, low level CSP infra services, turnup/deploy and break/fix automation
Running ML and GenAI systems not required but a nice to have.
We are looking for platform eng and SRE folks in CA and WA (various levels both IC and Mgr). DM me if you are interested.
BMaaS, Kubernetes, Slurm, low level CSP infra services, turnup/deploy and break/fix automation
Running ML and GenAI systems not required but a nice to have.
I'll be joining @nvidia as a Senior Director of Engineering for DGX Cloud next month!
How can GPU/TPU-as-a-service offerings improve for you no matter the CSP or even on-prem?
Agreed.
I'd add an insight could come with a confidence level (of how actionable it is).
Optionally, they could also be associated with the proactive alert categorization proposed in the following article by @StepanDavidovic (tech reviewed back then):
https://t.co/ZwpZP801K3
It’s so much of a cliché that it’s a cliché to write it down: software ate the world, but data is having it for dessert. Yet in my slice of the world, the dinner party is dealing with indigestion, not inspiration.
https://t.co/lhpqvZvFl1
"I have concepts of a plan" is half of tech CEOs. You know, the ones thinking that delegating some to seasoned managers and actually having C level reviews are mutually exclusive. I'm still in agreement with PG's footnote. This is all going to be misused.