​The ETL of Communication: Architecting Tech Talks Like Data Pipelines

Have you ever gone to a tech event, seen two or three talks, and at the end of the day, you couldn't remember a single thing the speakers talked about? Well, that is not your fault. Even if you tried to take notes or take a few photos of the presentation, if nothing stuck with you, the reason is usually the way the presentation was built.

​We, as technical people, have both a blessing and a curse: we know too much, and we want to share all the technical knowledge that we can. But how much of that actually makes sense to the people who are listening to us?

​I am Julia Molitzas, a Brazilian Data Engineer at Itaú Unibanco and an AWS User Group Leader. As a tech event organizer, I have seen all types of tech talks. I had the opportunity to participate in the AWS New Voices Program, and in this article, I want to explain to you why the way you present is just as important as the content you present.

​The AWS New Voices Program is a training for AWS Community Builders to learn about communicating and becoming better speakers. In a five week program, participants learn from an amazing speaker, Meridith Grundei, how to create a lightning talk, craft hooks for introductions, and most importantly, how to connect with the audience.

​In Data Engineering, we know that raw data is messy and practically useless for business users. If we just dump raw JSON files into a dashboard, no one can make a decision. The exact same rule applies to public speaking. We often step onto the stage and do a "Raw Data Dump", throwing complex architecture diagrams, blocks of code, and endless bullet points at the audience all at once.

​The most practical lesson I learned to fix this is simple: Tell them the one thing. Every single slide must have exactly one point. If you look at your screen and cannot immediately identify its core message, the slide is not ready. Instead of putting your entire script on the screen, filter the noise. The visual should only highlight the single takeaway you want the audience to remember.

​In a data pipeline, orchestration is everything, tasks must happen in a specific order to avoid failures. In a presentation, your slide transitions are your orchestration. Before this program, I used to click the slide, look at the screen, and then decide what to say. Now, I use a technique called Set it up. Before you even click to the next slide, you must tell your audience what they are about to see. This keeps you in control of the narrative and prevents the audience from being surprised by random visual shifts.

​Once the new slide is on the screen, you need to Anchor them. When we display heavy data or complex cloud diagrams, we assume people automatically know where to look. They do not. You have to verbally orient your audience. Simply saying, "Look at the left side of this diagram, where the data ingestion happens," completely changes how people consume your content.

​As technical professionals, we process information fast, so our natural tendency is to rush our speech to fit 10 minutes of content into a short lightning talk. But when you introduce a heavy visual, you must Give it space. You have to pause. Let the slide land and let the audience digest the architecture before you start talking over it.

​Finally, remember that Everything matches. What you say and what they see need to line up perfectly. If you are talking about data pipelines but your slide is showing a generic list of unrelated bullet points, you are competing with your own presentation. Humans cannot read a complex text and listen to a speaker at the same time.

​If there is only one takeaway you bring to your next tech talk, let it be this: Your audience's attention is a limited resource, so do not waste it on raw data.

​Organizing tech meetups has shown me that bridging the gap between deep technical knowledge and clear communication is one of the biggest challenges in our field. We often believe that presenting is an innate talent, meant only for extroverts. But it is not. Communication is an architecture.

​Just like we build reliable data pipelines with clear boundaries, data quality rules, and specific purposes, we can engineer our talks to deliver exactly what the audience needs. By applying these practical guardrails like finding the one point, setting it up, anchoring your visual, and giving it space, you stop dumping data and start actually communicating. Your message becomes the final, reliable product that your audience can truly consume.