Data Tools, Simplicity, and Smart Choices: A Conversation with David Ostrovsky, Software Engineer at Meta

At this year’s Software Architecture Summit during Bucharest Tech Week, we’re excited to welcome David Ostrovsky, Software Engineer at Meta and co-author of Pro Couchbase Server. With over 25 years of experience in large-scale distributed system architecture, David brings not only deep technical knowledge but also a passion for community and curiosity. 

His journey started young at the age of 9, a dusty library book sparked a lifelong fascination with computing. Today, David blends software engineering with speaking, writing, and training. In the interview below, he shares how this journey unfolded, and why DuckDB might be one of the most interesting tools in the analytics landscape right now. 

🧭 A Journey from Library Books to Meta 

David’s career began before most teenagers even consider internships. His first experience with tech, at Intel, introduced him to the world of data, an area that has defined his career ever since. 

“I landed my first “tech” job at 17, migrating data from one piece of proprietary software to another at Intel over the summer. That set the tone for the rest of my career: being the “data guy” wherever I went.” 

That early path turned into a lifelong focus, especially once the Big Data wave arrived. David was well-positioned to surf it. 

“I was lucky enough to jump on the Big Data trend in the early 2010s, which is what I’ve been doing ever since. Joining Meta was a pretty big milestone, of course. It’s fascinating to learn what happens after a company solves all the “common” data problems and finds itself tackling the sort of challenges that few others in the industry had to face before.” 

💬 Why He Keeps Speaking, Writing, and Sharing 

David doesn’t just write code; he writes, speaks, and teaches. For him, connecting with the tech community is more than a side project; it’s a source of energy and inspiration. 

“It’s fun! I love meeting people who share my interests and geeking out with them about technology. And I’m pretty sure I learn a lot more from the people I meet at conferences and meetups than they do from me.” 

🦆 DuckDB: A Simple but Powerful Data Engine 

When it comes to data tools, most engineers focus on scale. But David points to a powerful alternative that shines in its simplicity: DuckDB. 

“DuckDB fills a really interesting niche in the big data ecosystem, which very few other tools address. It’s an embedded analytical database that is highly optimized to work on a single machine. It pushes the boundaries of what we can do without turning to expensive and complex big data tools.” 

🔧 What DuckDB Does Well 

DuckDB isn’t trying to replace cloud-scale platforms but it’s perfect for when you need answers fast, without the setup. 

“The two most common scenarios where DuckDB excels are:  Fast, interactive exploration of data across multiple local and remote data sources. For example, running a bunch of analytical queries on a subset of a very large dataset locally on your laptop. This is especially useful if you are still not in a place where you can justify investing a lot of effort in setting up complex data architecture.  Quick-and-dirty ETL workloads, transforming and projecting data across all the common storage formats and technologies. Want to clean up and ingest a terabyte of CSV files from your clients into your data warehouse every night? Just run a single DuckDB command.” 

🧠 Choosing Tools Thoughtfully 

David’s advice on tools is refreshingly honest: don’t overengineer, and don’t reinvent what someone else already built. 

“Like every good engineering question, the answer to this one is “it depends”. But there are two broad principles I found useful in my experience:  The first is that I’m a big believer in the “buy” side of the build vs. buy tradeoff. If it’s not part of your core business, try to get it as a managed service, as long as economics allows. Don’t build a data warehouse, pay someone who already built one for you.  The second is that we often overestimate how much scale we’re going to need. Don’t build a big data platform until you actually have big data to put on it.” 

🛠️ Practical Advice for Simpler Data Workflows 

Sometimes, the smartest approach is to reverse-engineer from the business need. David shares a mindset that can help any engineer make clearer choices in data architecture. 

“…this is another one of those “it depends” questions. While it’s hard to generalize, I think the best strategy is to always understand the business impact of what we’re trying to achieve first. The thought process would look something like this:  We want to answer business question X In order to do that we need to be able to run analytical query Y In order to do that we need to create and manage dataset Z What is the minimal combination of tools and effort that allows us to do that reliably?  The third time this happens, we should also ask a fifth question: Can we generalize this into a shared solution, in order to make it easier to answer the next business question, and the one after that?” 

📌 Final Takeaway: Don’t Overbuild 

In a world where even our laptops can handle serious analytics, David leaves us with a reminder to keep our solutions proportionate. 

“The definition of what is “big data” is constantly shifting. What used to require running a cluster of Hadoop machines in your on-prem data center can now be done on your laptop or by giving your credit card number to a cloud service provider. So it’s always worth exploring the niche options before embarking on a big infrastructure project.” 

👥 Meet David Ostrovsky at Software Architecture Summit  

Whether you're working with analytics at scale or experimenting with simpler workflows, David’s talk will leave you with new ideas, a dose of pragmatism, and a better understanding of what tools like DuckDB can really do.  

📍 Join him live at Software Architecture Summit  

📅 June 20 | NORD Events Center by GlobalWorth   

🎟 Secure your seat today