Leonardo Schettini

Hi I'm Leo. A data scientist and software engineer for almost 10 years. But I'm more than that: I'm a cat parent, designer, curious, plant lover, confused, and more. I love to cook, paint, board games, research, bonsai, find new hobbies, travel, and more. And I keep evolving.

@RenaissanceSays @ea_yusi @threadreaderapp Whats the point of this unroll? It has ads just like x, and you still gotta scroll to read

@ea_yusi @threadreaderapp please unroll

This man was 20 years ahead. Early investor in Facebook, Twitter and Slack. Using each he grew his family business to a $60M business and built a media empire. Gary Vee's 6 predictions on social media for 2025 (and why you should care):

When learning languages I wanted some exercises related to subjects I work with. Duolingo-like exercises for example. Any tools out there for this? Ps.: even though chances are no one will need to learn new languages in the near future

@levelsio It was never alive

LangChain is dead

Anyway, it’s possible there are other ways to leverage this passion to the research stage. Does anything comes to your mind?

That’s not foolproof though. Paid reviews, AI generated content and etc could bias results. And implies people can actually trust whatever is used behind the scenes.

The first step to success is to master some specific skill. Like they say, top 1% on anything will be successful. To excel we need focus. Next, we need to leverage that to get what we want. But let’s take it slowly. 🧵

It’s about trusting your own words, and being truthful with yourself.

@GregKamradt Not years, but a few minutes it’s ok.

I would wait a year to hear its opinion on the perfect question to ask Then another three for the answer

@AndOchoa @robj3d3 @tryhygeia I’m up for it! I see some challenges like competing with no code tools and balancing being cheap vs customizable

@leoschettini2 @robj3d3 @tryhygeia That would actually be a great offer. Streamline the process to build all artifacts on a landing page to get buying intent. Pre-MVP Boilerplate Who wants to pair on that?

It's official! I'm launching an MVP development agency 🥳 And the first 2 spots are going for only $1,999 EACH! Here's the story of Liftoff.so: Liftoff was spawned out of a struggle. 11 months indie-hacking, and $0 earned 😅 Demoralizing to say the least. Having spent years leading engineering teams in startups and corporations, my skills were there. But my approach was wrong 🤔 I would build, burn-out, repeat. All without validating the problem at hand was actually something people wanted to pay for🤦 So, I'm trying something different. Bring my skills to you 🫵 Stuck with an idea? I'll help you make it a reality. First, we'll discuss your idea and create a detailed specification for your MVP 🎯 Next, watch your MVP take shape with regular updates 🔔 Finally, we'll deploy your product and walk you through the steps to ensure you have full confidence in running it on your own! 💪 So, are you ready for liftoff.so? 🚀

@robj3d3 @AndOchoa @tryhygeia Rob, you could also add that to your offer. Like building a page to collect leads, interest, pre payment… you could deliver that quite quick after you’ve done it a couple times. With some validation people would probably be happier to build the mvp or something bigger

@AndOchoa @tryhygeia Yes, generally best to come with an idea validated.

@AndOchoa @robj3d3 @tryhygeia Because in theory the MVP is supposed to be simple enough that you don’t need dedicated software. Like Amazon, ebay and similar could be validated nowadays with a whatsapp or fb group. That would be a nice exercise, thinking how simple you can go to validate an idea.

@robj3d3 Well I think they can do that without an MVP, for less money and faster. But I might be wrong. Me after one month with @tryhygeia 👇 x.com/AndOchoa/statu…

@asmartbear Ha, completely missed that was an article. Actually like you used “dead phrases” as a reason to what everybody suggests and do, like being specific I guess it’s about being relatable, and some stuff gets used so often they lose this ability

@leoschettini2 Yup, following my own advice from the article: That if you make fun of it, you're back to a good place!

If 78% of 'About Us' pages claim to be 'the leading provider,' then nobody is truly leading. How about: Be specific and actually say something real. Here’s how: longform.asmartbear.com/authentic-is-d…

Look not to fix your life into a better one, but to create the life you want! Because when we are creating we are taking action to make something be.

People doing social engineering with LLMs make me think what other parallels we can make between how we and AI think and behave.

@hamadamahdii Although not fully a guide, there’s the book zero to sold. And I’m sure there are others What is appealing from a guide/wiki?

Someone needs to create a rigorous guide/wiki to building a product geared towards #indiehackers & #buildinpublic folks. - Backed by at least some data and science. Not fully emotions and personal anecdotes. - Provides multiple approaches and delves into the nuance of each. - Not overly exaggerative. E.g. "Never offer free trials". - Doesn't 'sell' anything at the end. Just genuinely valuable info for people starting out. - Covers all of the areas involved in building a product. Also, the prerequisites needed to increase the chances of success. - In-depth case studies. - Open-source and allows for contributors. Nobody knows it all (even if they've made it and claim them do).

@hamadamahdii Interesting you building this! Following to see how it goes!

I keep making the same mistake (?) 😭 It all started with spending 2 hours reading through the Llama 3.1 paper. I've (temporarily) stopped working on Luminous to explore another idea. It went from... "Let me explore for a few hours" to working on it for 20 out of the past 48 hours 😂. We are definitely going to see more open-source LLM adoption. My hypothesis: lots of researchers/startups/companies will want to fine-tune these models. Starting with Llama 3.1 of course. It's the first open source model that has raised a few serious eyebrows. The Goal: To allow people to easily fine-tune Llama 3.1. Like, really really easily. No-code with a minimal barrier for entry. There are other services that allow you to fine-tune Llama 3.1. That's a good thing though. If no one else did it I'd be worried! A lot of them are geared towards enterprise. I'm hoping there will be more smaller players who will want to play around with Llama. I will make another post on my progress so far 😊. It's not fully ready yet but here's my landing page (credits to MagicUI). On another note: My wife is not happy with my lack of presence lol. I told her that it'll be worth it though. (I've said that before for another idea but it wasn't worth it! Let's see with this one 👀) #buildinpublic