Hello! Tired of the AI crazyness, there are solid approaches beyond deploying yet LLM to answer question or hook them up to a MCP. Do you have some time to spare and are looking for something cool to try out? Let’s try some computer vision model implementation to work out an open classification and detection problem, explore some non-llm models along the way and the approach taken to solve it, spoiler we’ll use YOLO + OpenCLIP for semantic video streaming search.

In this new era of AI-assisted tools, you and I are constantly being bombarded with news about the obsolete software developer. We are told we are no longer needed, that anyone can build a product, that we are disposable, that layoffs are everywhere, and that the train is leaving so we better jump before it is too late.

So now it’s friday I have already cool off a bit into what happened last week on April 2nd, I was taken out by the way of layoff and in the blink of a second stop being part of Automattic, I’m very proud of the work I did for the last 4+ years; so there’s not much else to say about it really, I have already made my peace with it, and I’m finally able to sit down, write about it, acknowledge it and move on. I’m very thanksful for the friends and colleagues that reached out to me and keep actively supporting me so let’s see what the future brings up next!.

Pozool is a point of sales aimed at small coffee/restaurant/truck food business, that back I developed back in 2014! Today I remember this time fondly and though to myself can we revive the same brand and tool and pivot it toward Short term rental business ?

Recently I’ve been going through the incredible book by Prof Elreath on statistical rethinking; it’s quite a good book to wrap your head around bayesian approach on causual analysis, as the book progress and starts to stack more and more concept together it reaches a point where a more sophisicated approach to sample the posterior probability will be required. Hence by the

Spacy es una herramienta para trabajar problemas relacionados con lenguaje, con un amplio soporte de lenguajes provee una sintaxis intuitiva para operar en ambitos linguisticos.

There are many phases when it come building a machine learning solution, from data preprocessing all the way to deployment, many of them with their sets of challenges and nuances, today let’s explore a simplistic approach to deploying a trained keras model that classify images of dogs and cats in a Flask application.

Recently I stumble with a pokemon dataset, and for anyone growing up in the 2000’s this brough up good memories, so of course I decided to take a peek into it. Let’s try to use logistic regression for predicting wether a pokemon would be a legendary based on features available, in the process we can look at how quickly explore the dataset with ggplot, splitting on training/testing, work with imbalance classes and looking at the result.

I recently posted an example on Text classification modelling with tidyverse, SVM vs Naivebayes, as usual when one is working with data unexpected things can happen, in this particular case it was a rather strange crash in RStudio that took me a couple of hours to figure out. TBH I think I spent as much time looking at this issue than the code for the post, but I digress. Let’s take a look at the isolated code that made R Studio kaboom:

It’s been a while since I wrote something, years actually, but here we go. Modelling is at the core of statistical learning, it allows us to make use of different techniques usually to predict, classify or find pattern within a particular dataset. USually the workflow involves a preprocessing with several tecniques and decision, such as imputation, omitting na, normalization, centering, and so on. Here I want to explore how the tidyverse has created an ecosystem that allows for fast and simple use of different modelling for comparison.

Well I have little to no time to write lately, but I still want to post something so here it goes a quick set of links for learning SQL (free of course). The relationship between SQL and data science is obvious, almost all companies have a relational database to store and extract data as part of their day-to-day operation, so is a must have. So here is the list of link to learn SQL

Let’s together explore another technique that it has several application which includes parameter estimation on mixtures models or hidden markov models, data clustering, and discovering hidden variables - EM Algorithm applications. To grasp the power of EM-Algorithm let’s considered an always familiar and simple linear models, and try to discover from which model was the points originated (clustering) with an iterative approach.

K-means is an incredibly simple algorithm, with a concepts that can easily be explained to anyone. But in order to understand something better sometimes is nice to build it from scratch. R and python both have some amazing libraries for clustering, but the beauty of K-means lies in the fact in that is so simple that its implementation can fit into a tutorial like this one.

An easy look into animation with R by simulating Brownian motion and creating a GIF in a few steps. This post is based on Javi Fernandez Post “Brownian Motion GIF with R and ImageMagick”.

So are you exited about all the news about machine learning you’ve heard lately? I sure was a year ago, and it does not take too much times to find a significant amount of information to start with. Fortunately now there is a lot to start so I wanted to make a list of videos, articles, books that I’ve found to be incredible useful to start with machine learning.

Let’s talk Artificial intelligence, not only machine learning. I wanted to make an instructive manual on different AI content that I’ve come across in the forms of articles, concepts, practical examples, tutorial or trending news related to it. I’ll try to brief and concrete as possible without water down some important concepts, thanks for reading, see you soon.