README Post

Welcome to this corner of the Internet, and to this very first post, where we will be exploring the motivation behind the genesis of this blog, and defining the road map toward future projects (hence, future posts!).

Why this blog?

tldr; Writing is thinking. I learn a lot while writing. Public learning has the potential of exchanging ideas of topics I care about, and connecting with interesting people while at it.

That's the short version. The longer version is an origin story.

As described on the home page, I am a computer scientist (student, as I am one subject away from the graduate degree) with a major in Artificial Intelligence. I have decided to specialize myself in AI as part of a career shift that took place 5 years ago. At that time, I was studying economic sciences, more specifically, actuarial sciences. I had a phenomenal statistics lecturer who made me realize that data processing and unsupervised pattern inference was truly just taking off, and that the potential of forecasting models based on machine learning was huge.

I started reading, and learning more about Artificial Intelligence. This was the first time I had to deal with computer programming concepts, and also the moment I began to understand the usefulness of abstraction in information systems, the difference between difficulty and complexity, and how the latter can scale up in the real world to levels we could never parse in a lifetime, but still can be understood, and sometimes even predicted, with the right tools and techniques.

Intriguing stuff that, by the age of 20, made me truly question my previous career decision. Economics was interesting on its own (I might even write a more philosophical post about that later), and I had plans to graduate and start working in finance. But then, every time I was writing a formula in Excel, or using an SAP GUI to complete my tasks, I couldn't help but notice that I had no clue what was going on behind the screen. I just felt that I lacked the depth of understanding about information systems that I considered necessary to grasp the real value of resources in today's economies.

Between all this brain-noise, conveniently hapening during the COVID lock down, and the very effective targeted marketing from e-learning platforms, I started doing R and Python courses. Later came the degree switch to CS, and just like domino pieces falling one after each other, here we are today. Thanks Udemy and FacebookAds I guess. Long story short, computers and I became friends now, and I find myself writing this Markdown file on a Vim-based IDE for my over-engineered NextJS blog.

The Artifact

So, that's the origin story. Now that I (almost) got my CS degree and feel confident with my understanding of information systems, I want to start exploring other domains I always found interesting, such as finance/economics applications, where I could leverage all the knowledge and tools acquired. For that, I have a plan. But just building a portfolio of projects feels incomplete. In a world where generative AI can produce perfectly working programs, the process (the human thinking behind the code) is what matters. This blog is my way of showing that process. Writing, for me, is a form of thinking that connects my mind across different moments in time. It forces me to design a system: gather external information and ideas, process them, and produce a synthesis. This blog is the artifact that serves as an interface to the human story behind the repositories. It's an approach I found for gamifying this career path, generating an opportunity to connect with people, and showing my value as a software developer.

What to expect?

What is the plan then? I will be writing two different kinds of posts. The first type belongs to the main quest, landing a software development / engineering job, ideally in "AI Engineering". I want to position myself at the intersection of System Architecture (integration of AI in data pipelines), and Machine Learning engineering (designing, training, and deploying ML models). I want to keep working as a software developer (whatever that looks like in the future) while at the same time gaining understanding of the value of data and its processing. During my degree I noticed that one of the best ways to learn not just the theory but the best practices and nuances of anything is by working at it. Following this principle, the main quest consists of building three portfolio projects with the goal of finding a job that will keep feeding the virtuous learning circle. I didn't pick these projects at random: I looked for a domain that would force me to deal with high complexity, real-time data, economic utility, in which performance is quantifiable.

I selected the financial domain as my sandbox for now. Not necessarily because I aim to work in Fintech (although it would be quite a cool experience), but because I consider the constraints here are useful for training:

1. Complexity of dealing with messy, heterogeneous data.
2. The need of designing robust and scalable architectures to handle the huge time-series. Such agentic pipelines are not trivial to build, and the reasoning will probably be transferable to many other domains. This does not mean, future projects will necessarily follow this trend as I am curious about numerous topics and domains of knowledge.

That leads me to the second type of posts, which will correspond to "side" quests. These are more about fulfilling my curiosity. It happens to me frequently that I am working on an assignment from work or university and I notice a slightly unrelated topic that wants a piece of my attention, but the context is not the right one, so I write it down for later. Well, this would be the right context for those notes. Side quests won't really have a consistent theme or structure. So, if you are indeed a recruiter learning about me, I'd recommend you just stick to the first type of posts. I'll make sure they are tagged properly ;)

Defining the main quest

In order to define these main quest projects, I did some research on LinkedIn. I went through many job posts, looking for a sample of >30 job descriptions I liked (mostly AI/ML Engineer), ranging from startups to established tech firms. From those, I extracted the requirements and computed some statistics to gain useful insights.

Considering I worked with most of these technologies, and studied the theoretical concepts as well, I designed a series of projects that would allow me to actually show that (along with their corresponding blog posts!). The projects are also devised to construct a bridge between "building software" and "training models". My mini-research showed a clear path: Python was always present, Cloud Infrastructure/Docker and SQL came in second (~80% of the job posts), and the ability to build Agentic Workflows / RAG was a key requirement in most jobs I found interesting. With that in mind, I challenged myself to do the following projects:

Project	Title	Description
1	Financial Data API	The foundation. A robust Python backend (FastAPI/SQL/Docker) to ingest, clean, and serve financial data. It solves the "messy data" problem.
2	AI Financial Analyst	The integration. A Next.js Dashboard consuming the API from Project 1. It integrates LLM Agents (RAG) to interpret the data and generate reports, flexing some user-facing AI.
3	Generative Market Modeling	The model. Moving beyond APIs, in this project I will use deep learning (VAEs/Transformers) to model market dynamics and generate synthetic scenarios. Hopefully with some utility.

About the format

Most posts will be written in a free long format. Similarly to this one. There is always a README file in every repo for structured, efficient, and functional communication. We are here to have some fun and understand ourselves, humans behind keyboards, not the programs.

The next post (I have already started working on it) should be about the data API made with Python.