We are excited to announce the second night of talks in the NYC Systems series! Talks are agnostic of language, framework, operating system, etc. And they are focused on engineering challenges, not product pitches.
We are pleased to have Ben Hindman and Stephanie Wang speak, and glad to have Trail of Bits as a partner for the venue.
Co-creator of Apache Mesos. Co-founder of Mesosphere and reboot.dev. Early Twitter engineer. Inaugural member of the Technical Oversight Committee (TOC) for the CNCF.
Over the last two decades, backends have evolved from simple monoliths to complex distributed systems, incorporating microservices, serverless functions, multiple databases, event queues, and more. These architectures improve scalability and modularity at the cost of significant complexity in managing data consistency.
Even without the extra complications in the backend, the inherently distributed nature of the frontend and the backend was already complicated enough.
We’ll present a new programming model for building full-stack distributed systems, Reboot, that unifies application logic with data management. The programming model relies on built-in primitives for reactivity, atomicity, and idempotency that dramatically simplify development without sacrificing safety or scalability.
Instead of thinking in tables and rows in a database, developers think in data structures and functions, and the programming model provides a well-defined semantics for how those functions can be composed safely together.
Stephanie is a Staff Engineer at MongoDB, where she focuses on designing and building scalable, high-performance database systems. Previously, she was a founding engineer at MotherDuck, where she played a key role in shaping its cloud-native analytics platform built on DuckDB. Prior to that, she worked on Google BigQuery, specializing in BigQuery client libraries and connectors.
Stephanie is passionate about developing solutions that enhance performance, scalability, and usability. She is particularly interested in the challenges of cloud-native data systems and ensuring that complex infrastructure remains robust and efficient.
You’ve got 100K documents and a mandate to “do AI”. Do you need a real-time distributed vector database on Kubernetes? Probably not.
This talk is about the systems we overbuild, the complexity we inherit, and why most problems don’t need distributed infrastructure – at least not yet. Through the lens of semantic search (and some spicy anti-patterns), we’ll explore how to scale right: by building a single-node system so good, scaling out later isn’t a rewrite, it’s just a rollout.
Because sometimes the fastest way to go big… is to start small and actually get it right.