Drew Koszewnik

About Drew Koszewnik

High impact software development professional with experience in performance-critical large scale web applications. Demonstrated ability to influence a team strategy and prioritization which achieves goals aligned with an organization\'s overall strategy - and to galvanize fellow team members to execute on that strategy. Possesses a unique ability to invent original solutions to complex technical issues. Focused on performance; able to quickly determine near-optimal algorithms to address specific problems. Passionate about spreading knowledge across teams and roles. Strong problem-solving, communication, and teamwork skills.

Experience

1. Principal Software Engineer (E7)

   Netflix

   Jul 2022 — Present · Los Gatos, CA, US

   The highlight of my tenure in this role has been the creation of RAW Hollow (“Read-After-Write”), an evolution of the NetflixOSS Hollow technology I had previously developed and open-sourced. While Hollow enabled large-scale, immutable in-memory datasets for high-throughput services, RH extends the model to deliver real-time, read-after-write consistency across a distributed system.Partnering with Eduardo Ramirez and Govind Venkatraman Krishnan on this work, we built RH on the consistent observation that co-locating data in memory yields transformative engineering benefits: lower infrastructure costs, simpler codebases with less tech debt, more robust services with reduced downtime and improved performance, and greater development agility for business-critical initiatives.RH has been in production for a year, hosting 100+ production datasets across diverse Netflix use cases - including mission-critical, tier-0 (“must not fail”) systems. It is the fastest-growing data abstraction in Netflix history. I genuinely believe the innovation behind this technology has the potential to change how the industry approaches building distributed software, by enabling teams to reason about global state as easily as local memory.Another highlight was designing and implementing a novel technique for computing and communicating subsets of the Netflix content catalog - from zero to millions of titles - with orders-of-magnitude improvements in performance. By dramatically reducing the cost of constructing these sets and the message size between services, it\'s now possible to perform highly scalable, remote queries across the full catalog more efficiently than previous local in-memory approaches. It has since become a foundational component of Netflix’s content merchandising strategy.Alongside this work, I have contributed to initiatives across the content, platform, and member organizations, providing technical leadership, architectural guidance, and hands-on engineering.

Education

• Lawrence Technological University

  Bachelor of Science, Computer Science

  2001 — 2005

Skills

• Linux
• Lucene
• Junit
• Cvs
• Solr
• Cassandra
• Tdd
• Relational Databases
• Unix
• Agile
• Test Driven Development
• Eclipse
• Ant
• Software Development
• Nosql
• Algorithm Design
• Maven
• Rest
• Subversion
• Web Services
• Distributed Systems
• Java Enterprise Edition
• Teamcity
• Soap
• Jboss Application Server
• Tomcat
• Integration Testing
• Unit Testing
• Cruisecontrol
• J2ee
• Hudson
• Jboss
• Agile Methodologies
• Information Retrieval
• Amazon Ec2
• Data Mining
• Perl
• Java