brunooliv

Small weekend project, using smolagents to build a simple agentic flow that can generate and execute DB queries from plain English!

There’s a big difference when someone stresses over minutiae and fails to see the bigger picture or why the overall design might be flawed. Wanting to have code up to a given standard is a laudable goal but as with everything, it can be taken to such an extreme that people lose sight of the real value being delivered. Code is not the end goal, except for some people it is.

Wow, this was so amazing! And in a way, so... refreshing! Makes me want to learn Ruby :D

I found strange to see so much criticism of the post as usually, anything that Wes Kao writes really resonates with me as a senior IC. Usually there’s a great balance of nuance and depth to the advice that makes it easy to digest and pick small bits and pieces to try out on my own context

This assumes that human reviews also catch these which is DEFINITELY not the case either.

As long as you have good pipelines, linters, a careful suite of tests at different levels like unit, integration, e2e and if you can test things in an acceptable like environment then human code reviews offer very very little benefit…

This is a post that just reads as if the author is still in the “honeymoon” stage of their career where programming is seen as this extremely liberating and highly creative endeavour that no other mortal can comprehend. I get the feeling and I was there too, but, writing code has always been a means to an end which is to deliver business value. Painting it as this almost abstract creative process is just… not true. While there are many ways to attack a given problem, the truth is once you factor in efficiency, correctness, requirements and the patterns your team uses then the search space of acceptable implementations reduces a lot. Learn a couple of design patterns, read a couple of blogs and chat with your team and that’s all you need. Letting an LLM write down the correct and specific ideas you tell it to based on what I wrote earlier means your free time to do code reviews, attend important meetings, align on high level aspects and help your team members all which multiply the value you deliver only through code. Let LLMs automate the code writing so I can finally program in peace, I say!

This post is bookmarked for me since this first time I read it and I occasionally come back to it, it’s a great one

I think that the influence of 100 engineers is easy to answer: if you set technical direction as an EM or VP of engineering or whatever level E8 is, this means that _indirectly_ many teams will scramble to work to fullfill the exact technical direction you set, which means in turn that managers and senior engineers in those teams will be working on something that you at least had a saying in. That’s how you affect 100s of engineers.

More directly if at some point a bug appears and it’s a blocking issue for releases over an entire department or vertical in the company, if you fix it, you can even affect SEVERAL hundreds of engineers. Apart from these very high level overviews I doubt that E8 folks will look at a chunk of work and start thinking about the number of engineers they’ll affect… It’s also much more important at that level to be truly product oriented and seeing where the real impact can lie. Additionally to this, don’t forget the classic “promotion driven development” where features are reinvented or “trivial” changes that affect an entire user base are rolled out, etc

Location: Amsterdam , NL

Remote: yes, or hybrid

Willing to relocate: No

Technologies: Java, Springboot, Python

Résumé/CV: https://nl.linkedin.com/in/bruno-oliveira-78181176

Email: olivbruno8 @ gmail . com

Great post thanks for sharing this “from the trenches” it’s always a joy to read these deep dives! Thanks

I think while the basic idea of advocating for stack consistency between the main project and any support scripts is very nice, I always find it not worthy to pursue in practice for several reasons: a lot of the “environment” around the main codebase will be a weird mix of YAML, bash and a scripting language like Python or Ruby for things like gitlab, airflow, GitHub actions, etc. Given this heterogeneity of the project environment and additional complexity of “forcing” something to use a language it might not be very well suited for, makes this really a no brainer for me: use the most convenient tool for the job. Plus as a JVM bound developer I love me my occasional Python

Played with several LLMs to create some cool fusion recipes! New on every refresh! Get inspired and cook new stuff

Is Claude 3 on a pay per use case or a fixed monthly fee? I’ve used it a bit with the free 5USD they have and it’s looking so so much better than gpt 4… the problem is I have the feeling it’s also extremely more expensive

Do the implications at a practical level mean that the size of gguf files will become smaller?

Damn, I see, how many tokens per sec you get on that setup?

On a Macbook M2 I get ~10/12t/sec which is a tiny tad bit too slow for continued/ daily use, but if I think its worthy I might invest on a more powerful machine soon-ish!

This is really great news, I will definitely be on the lookout for it :D

Mixtral aka the 8x7B the "sparse mixture of experts" one is not the same as, eg. Mistral-7B which is still very, very good, just not quite hitting the mark on some things.

I still couldn't run Mixtral 8x7B on an M1 Macbook Pro with 32Gb ram, so maybe I am indeed doing it wrong? Or are there better quantized versions available now or..?

Anyone who uses these models for more than 10 min will immediately realize that they're really, really bad compared to other free, OSS models. Even Phi-2 was giving me "on par" results except that its a model of a different league.

Many models are being released now, which is good to keep OpenAI on their toes and not mess up, but, truth be told, I've yet to see _any_ OSS model that I can run on my machine being as good as ChatGPT 3 (not 3.5, not 4, but the original one from when everyone went crazy).

My hopes for consumer hardware ChatGPT-3.5 within 2024 probably lie with what Meta will keep building upon.

Google was great, once. Now, they're a mere bystander in the larger scheme of things. I think that's a good thing. Everything in the world is cyclic and ephemeral and Google enjoyed their time while it lasted, but, newer and better things are and will, keep on coming.

PS: Completely unrelated, but, gmail is now the only Google product I actively use. I don't, genuinely, remember the last time I did a Google Search... When I need to do my own digging I use Phind these days.

Times are changing and that's great for tech and future generations joining the field and workforce!

I’d read a book in this style if one was written, Xe! This is both entertaining and very well written

This book ALSO changed my life when I was first learning programming!! Not so much in terms of a pivot but of the way the book was written itself and how some concepts all of a sudden just made sense as a total freshman. Now, ~10-15 years later, seeing this third edition is such a nostalgia and I can't recommend this book enough!!!

I find this super weird and almost borderline invasion of privacy? I mean, a job is your professional life and you’re there to work, not go directly make friends or stalk people… I mean sure I’ve made a few people whom I’d call friends in previous jobs and current one too and I’d like to believe that we’d have enough confidence in the friendship to tell each other about quitting. But seeing that potential info about anyone feels very weird…

This is awesome! A question for Italians in the audience though: is using butter as a fat something that’s commonly done? Being Portuguese and fond of Mediterranean cuisine, I always saw butter as a fat as being a more American thing but maybe that’s just the wrong perception and it’s totally fine if you use a dab of butter? PS: I do it myself, reason being that usually I’ll use extra virgin olive oil while preparing a sauce and I like to add butter to save on the oil, which can be quite expensive.

It’s all super basic and the recipes are hardcoded in the JS side and you can only search for one ingredient at a time. I WANTED this to be simple and did this while in between some downtime all on my phone to see what I could push with ChatGPT. Turns out quite a lot. 2000 or 3000 recipes is an amount that nobody ever will cook all at once. Many more still will be repeated, etc. This all fits PERFECTLY client side and leveraging local storage for the favorites is very doable. I wanted to combine both usability and simplicity. Maybe later I’ll support searching by multiple ingredients.

The idea truly was basic simplicity here

Hey @whycome thanks for the feedback here it is: https://impossible-emphasized-ash.glitch.me/

Now responsive!

Will do!! Thanks!

It’s already happening… it’s a total mess of weird microservices that make no sense to reason about and the hidden costs on things like free tier etc… not good

Thank you so much!! Really glad you liked it!

Obviously!!!

Additionally and if it's not too much to ask, a review on Amazon could really help me too! I'll definitely ship you a free copy to Poland once I fix the page numbers