Tacos de Datos
Spanish-language newsletter and community focused on AI and data. 20K+ followers, 1100+ active members. Weekly posts on AI Ops, practical implementations, and emerging tools for the Latin American tech community. Building accessible technical content where language isn't a barrier.
Technology Stack
Links
Frequently Asked Questions
How did you start Tacos de Datos and grow it to 20K+ followers?
I started Tacos de Datos in February 2018 over a weekend, wanting to create a blog where I could share cool data stuff I was learning—in Spanish. In 2018, everyone wanted to be a data scientist, but everything was in English. I wanted to show that small automations, like manipulating Excel with Python, can free up hours for many people. You don't need to learn English just to read a 400-word blog about opening Excel files with Python. I created a Twitter account and started sharing. People resonated with the content. I invited others to post, started a podcast with interviews, and created courses, collaborating with Código Facilito on bootcamps. My strategy was simple: be honest, be myself, and not take things too seriously. Tacos de Datos is meant to be accessible and demystify data science. Consistency is what has taken us further—the times I wasn't consistent were when we grew the least.
What's the biggest challenge in building a Spanish-language technical community?
The biggest challenge has always been consistency, because I also have a job and other side projects. Finding what people care about is difficult. Community management is challenging because I want this place accessible for everyone—whether you've been doing computer science for 15 years or 15 days, whether you've never touched a terminal or only work in Excel. Computer science is mostly male-dominated, highly technical, with few soft skills, and it can get rough. We have zero tolerance for that. Making it easy for beginners to ask questions without fear of ridicule is crucial. Lastly, balancing highly technical content with intro-level material is challenging. There's lots of intro content, some niche high-level deep content, but that medium tier—what we need to grow in Latin America—is the hardest balance to find.
What content resonates most with your audience?
The content that resonates most is practical advice—tutorials that actually work where people can copy and paste code, clone the repo, and get it running. That gets the first layer of engagement. What has the biggest impact, however, are reflective, philosophical stories that explain and demystify difficult concepts. For example, describing linear regression in terms non-math, non-technical people can understand, or breaking down an algorithm that predicts lottery odds into simple instructions and average-case probabilities. Those layperson explanations have the greatest effect. People who have followed Tacos de Datos for a while tell me these are the most impactful on their lives.
What advice would you give someone wanting to build a technical community?
Being consistent is the most important thing. The hardest part is not getting there; it's staying there. Be consistent and be kind to yourself, because it's really consuming. It is important. Lastly, don't do it for the money because there's very little money in it—you'll be losing money. Find other ways to measure your success and the things you care about. Early on, defining your success is very important.
Where do you see Tacos de Datos going in the next few years?
The next step is a learning platform we've been wanting to do since the beginning. I've been building it—we're building in public at enparallel.org. It's a new way to learn. We're curating courses with videos and presentations, using AI to tutor and explain things. Each presentation has a "help me understand" button where you can chat with an LLM that understands your context and explains in whatever way makes sense to you. We're also experimenting with course creation—following proven guidelines, you can create a course like "teach about vector databases to senior data engineers, six episodes," and the platform generates it for you with all our pedagogical features.