From Building NYSE Trading Technology to a Data Streaming OS

Описание: Why would the engineers who built the New York Stock Exchange's trading engine — processing $100B+ in daily transactions — leave to start a data streaming company? In this first episode of Inside the Data Plane, AlgoX2 CEO Alexei Lebedev sits down with co-founder George Levin to tell the origin story.

Alexei traces the journey from his early days in high-frequency trading, where he first encountered the chaos of distributed computing — events arriving out of order across data centers, state scattered across servers. That experience led to building the transaction engine behind the NYSE, using journal-based architecture and formal methods to guarantee zero data loss at exchange speed. The same approach now powers AlgoX2.

In this episode, you'll learn:
• How an exchange outage (and an Excel spreadsheet) inspired a new architecture
• Why journal-based systems are the foundation for reliable data — from file systems to blockchains
• What a "data streaming operating system" actually means
• The two-layer model: sequencing chaos into order, then executing deterministically
• Why Alexei argues blockchain is actually centralized, not decentralized
• How AI is making data streaming infrastructure more critical than ever

If you're building data infrastructure, running Kafka at scale, or just curious about how the NYSE actually works under the hood, this episode is worth 17 minutes of your time.

🕐 Timestamps
0:00 Introduction — Meet the hosts
0:54 The origin story begins
1:12 High-frequency trading and the chaos of distributed computing
3:07 From HFT to building a financial exchange
3:30 The outage that started it all
4:24 Journal-based architecture — from NTFS to NYSE
5:22 How do you convince the NYSE to adopt new technology?
6:10 Formal methods and correctness by construction
6:57 OpenACR — the code generator behind 1.6M lines of code
8:11 Why Alexei left NYSE to start AlgoX2
9:10 What is AlgoX2? The data streaming operating system explained
10:05 Fan-out: one message, ten consumers — why it's hard
11:07 The two-layer model — ordering events, then executing deterministically
12:26 Blockchain is actually centralized — a contrarian take
13:39 Layer one: why sequencing matters more than you think
14:42 Data streaming as a control plane
15:17 AI, clusters, and why nothing fits one computer anymore
16:40 What's next — Episode 2 preview

#DataStreaming #DistributedSystems #systemdesign
https://www.algox2.com