WorldQuant

World Champion Water Polo Player Miklós Gór-Nagy ’s journey presents what it really takes to perform under pressure. His perspective on discipline and focus is exactly the mindset we hope to see among the participants of IQC.

Most Spark pipelines don't need Spark anymore. Last year DuckDB ran a 265GB, 6B rows benchmark on a 2012 MacBook Pro, and all 22 queries completed on a laptop that sells for $80 on eBay today. Going distributed wasn't wrong in 2012 - spinning disks and 4GB of RAM made a 100GB dataset a real infrastructure problem, and distributed was often the only path through it. The issue is that hardware kept moving while the playbook stayed the same. Single-box RAM hit terabytes, SSDs got fast enough to do what clusters used to do, and a Fivetran study of Redshift and Snowflake found that the median query reads 100 MB while the 99.9th percentile still reads under 300GB. Choosing Spark for a 200GB dataset in 2026 should need the same written justification as choosing Kafka for 3 event per minute - the case might hold up, but defaulting to distributed without writing it down is where the lost decade came from.

It was a pleasure to meet Mike Bloomberg in New York for a thoughtful discussion. It was valuable to exchange ideas not only on the latest in technology and innovation, but also on broader philanthropic efforts, including expanding access to high-quality education and using data and AI to help solve complex problems.

I’m pleased to share Jordana Upton and my contribution to this year’s Milken Institute #PowerofIdeas series. We argue that in the age of AI, leaders must act less like traditional decision-makers and more like “systems architects” – building trust, ensuring diversity of thought and designing the conditions where refined answers can emerge from data and insight. We talk about the evolution of what an ideal leader has looked like through other technological advances and what that means in today’s environment. Looking forward to continuing the conversation at #MIGlobal in the weeks ahead. Read more here: https://lnkd.in/e37xSvww

Data engineering is going through what software engineering went through 15 years ago - the move from "it works on my machine" to CI/CD, version control, code review, automated testing. SWE figured this out around 2010, and DE is only now catching up. You can see it in how juniors react to legacy ETL tools like SSIS, Informatica, Talend. They don't want to touch them, and most people read that as "juniors want shiny things." I think it's something else. These tools made real engineering practices impossible - you can't code-review a drag-and-drop flow, you can't unit test it, deployment means clicking through a wizard, and onboarding means "ask Dave, he built it three years ago". Juniors learned git, wrote tests in school, ran CI pipelines - and then they show up to a job where the data team's workflow looks like 2008. Of course they push back. Companies still running GUI-based ETL in 2026 aren't making a technology choice. They're choosing an engineering culture where nobody can review each other's work, test before deploying, or trace what changed last Tuesday. The shift from drag-and-drop to code-first pipelines isn't about old vs new - it's about whether data teams get to work like engineers or not.

Bringing lessons from Olympians and World Champions into the world of quant. The Champions Journey — ambition, discipline, resilience — is the same mindset that powers success at the International Quant Championship. Patty Solimene Miklós Gór-Nagy Cristina Teuscher, OLY @Saina nehwal