akhayam

I have been waiting for an AI powered tutor to teach my ~10 yr old coding and writing, so this just made my day. Sal Khan and his team are incredible and I can't wait to start trying out Khan Amigo.

I posted this question without an opinion because I have a strong bias that it doesn’t matter when you start coding in your youth. I had no access to computing until I was in 11th grade. I feel that it helped me build better mental models because reasoning and build better mental a that I wouldn’t have built if I was younger.

Appreciate the community giving their perspectives, most which are spot on.

Nah... Mendable seems to be searching in documentation. This idea is to extract an execution model of every component in your system from the logs of that component. We can think of the prediction step as a complete separate machine model for now.

Anything else comes to mind?

Sounds cool, but how is this different from Colly: https://github.com/gocolly/colly?

I am not advocating to remove the foundations. My stance is that it shouldn’t be the first exposure to programming. I am all for covering foundations in later courses.

I agree that languages close to hardware are the wrong starting point to introduce a newbie to programming. And I see a change happening already—e.g. Stanford changing the introductory programming course to JavaScript in 2017: https://stanforddaily.com/2017/02/28/cs-department-updates-i.... I suspect that other top schools will follow suite whenever someone has the will and courage to design a new course in a higher level language. It’s a lot of work and the incentives for university professors aren’t aligned with making such a big investment into something that isn’t glamorous. In other words, don’t hold your breath.

Wow... that was quick. Would love to see what results you get.

Great to see Amazon opening and office in Iasi. I had my Frontend team there in my past life and it was top tier talent.

Would strongly recommend this little known jewel to anyone who is considering an offshore office. Not as cheap as some other locations further down east, but the quality bar is high and the city’s got its own micro culture which is very tech centric.

Seriously… “reset the drive” is still a thing?

“Do you pine for the days when men were men and wrote their own device drivers?” https://en.m.wikiquote.org/wiki/Linus_Torvalds

This was such a great read and took me back to the days when you had to figure out the right pin configuration on your modems and write device drivers for anything that was a bit fancy on Slackware.

Thanks for the nostalgia.

Gotcha! That makes sense. I would recommend looking at LangChain though, as it does a good job at modeling multi-stage learning / inference environments.

I believe AWS struck the right balance to "Think Big" and not worry about minutiae at the beginning, while tightening the belt and ensuring costs are not out of control at later stages. One of my teams was given the charter to map the entire world's Internet weather map. This service collected billions of data points every 5 mins and we spent many 10s of millions doing it. But as soon as the innovation phase was over, I had to take a goal and start an MBR (Monthly Business Review) to drive the costs down. This was a really interesting mechanism which doesn't stifle innovation in the short-term but doesn't let costs go crazy in the long-term.

Did you see a similar pattern in EC2?

It should exist and would be super powerful considering all the recent advancements in language ML. Here was the mental model of my model: the canonical representation (i.e. a representation after taking out run-time populated fields) of a log line represents the smallest meaningful unit of this "log language": _a word_. Taking this analogy further, an event is a collection of logs that occur together (mostly in order)—just like words spoken together form _a sentence_. Finally, collections of events that occur in close proximity (in time) represent _paragraphs_, while paragraphs occurring in a certain order constitute _chapters_. Using this mental model opens the door to apply all the new AI techniques for text extraction, summarization and generation to extract the semantic structure of any "log language" and then learn and classify behaviors observed at run-time. The eventual objective function is not generation though--it's reasoning with the optimal FP-TP tradeoff on a ROC curve.

I haven't seen anyone do it yet. Maybe companies like Splunk and Elastic will take a lead here. I am happy to engage, advise and contribute if there is an open source project around this. Has anyone else seen something remotely close to this?

Don't think I quite follow. Is the new program (operating on the output of the earlier program) supposed to reason about why you are seeing the result that you are seeing? Or is it doing more post processing to make the earlier output directly consumable by your corporate systems.

About 8 years back, I was leading an engineering team which was the escalation path for customer support. We were sitting on a large corpus of support tickets but didn't have any insights. I was amazed when word2vec came out and blew my mind. So I built a language model that trained on support ticket data. I modeled system logs attached to support tickets as an NLP model to predict what was going to fail next and for which customer.

Never made it to prod but was a great tool for me to see where I want to budget my team's time.

This is way before all the LLM and Generator models, but it was such a fun project.

Yes, there seems to be a design flaw here. There should be no SPOF in region design.

Water intrusion?

Is it raining a lot in Paris?

Etcd is what holds k8s together. I saw the concept of watchers evolve from what-used-to-considered a fringe use case to something so fundamental and powerful in k8s.

The only issue is that its usage best practices are not followed. Most users don’t back up etcd, using it like persistent storage. Resultantly, in a few upgrade cycles, nobody really knows what’s store in the etcd store and it becomes a “super pet”.

This article does a good job laying out the best practices. I hope users start adopting these too.

It’s slowed down for sure since COVID. Having said that, RSA is happening in SF and is heavily attended right now. There was also a huge meet up around Generative AI a couple of months back: https://sfstandard.com/technology/techies-swarm-san-francisc....

Try searching for a specific topic of interest and I am sure you’ll find some good meetups around that topic in or around the bay.

This appears to be introducing a math prerequisite for graduation. Not sure if there is anything deeper though?

Step in the right direction, but guidance alone won’t cut it. We need stricter enforcement of these guidelines as well. Not sure if that’s the VC’s job though.

“AWS cloud unit was generating operating profits of more than $2 billion a quarter when at a similar revenue level.”

I was at AWS when this was unfolding and cost was never an afterthought. I also felt that the best way to control costs is to make it a section in design docs. This would force the developers to consider costs when proposing design alternatives for a new service or enhancing an existing service.

That makes a lot of sense. Thanks for clarifying.

Hope my questions help you in making an sharper elevator pitch to customers.

Hmmm… I guess this puts Code Preview in the Netlify / Vercel camp… although those projects are richer in their feature sets around deployment tooling.

Is that a fair comparison?

Agree that you are not behind. Having said that, I am generally a believer of the "10K hours rule" to get really really good at something: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4662388/.

So you better be ready to put in those hours.

Sounds interesting, but what's the backend for Code Preview? and how is it different from other companies in this space like Retool, Zeet, etc.

Wow! I had no idea that the idea of an Atom goes back 2000+ years. Thanks for sharing this terrific post.

Goes to show how these tiny, simple things come together with some structure to create everything in our universe.

Andy Pavlo is incredible. I studied EE and have learned about databases through his lectures.

Here is the Tech Crunch article: https://techcrunch.com/2023/04/11/twitter-inc-is-now-x-corp/.

SSDD as far as I can tell.

Newbie error. Didn't realize that was the norm of the community. Tagging all older posts with the year now.