Website Netflix
Under the banner of Open Connect, Netflix operates over 10 thousand dedicated Open Connect Appliance (OCA) servers worldwide, serving content across billions of TCP connections per day. Every Netflix customer’s streaming QoE (quality of experience) depends in part on OCA servers rapidly adapting their transport connections to prevailing network conditions. The OCA Transport team is at the forefront of our advanced transport protocol design, implementation and testing. With every OCA able to log traffic statistics down to per-packet granularity, we have access to terabytes of amazingly detailed insights into the behaviours of real-world IP networks. And our team’s design choices have a worldwide impact on people’s enjoyment of Netflix.
You will lead and own the development and refinement of our best-in-class network lab environment for controlled streaming transport experiments, and work closely with transport protocol, client streaming engine and data analytics colleagues to ensure our lab-based experiments deliver actionable insights. You have a track record of designing, refining and implementing practical solutions for measuring IP network behaviour and using that knowledge to improve the performance and robustness of existing IP-based transport and networked-application protocols. You are excited by the prospect of working on the challenge of streaming content at-scale with high QoE over the diverse capabilities of service provider networks world-wide.
You will succeed in this role because you have:
- The ability to define and self-manage cross-functional projects fueled by ambiguous questions. You are a self-starter, curious and not afraid to ask when in doubt
- Passionate familiarity with how dynamic TCP connection behaviours impact the performance of networked applications (HTTP-based content streaming, …)
- Intimate familiarity with identifying and measuring issues impacting end to end IP network performance (such as bottleneck cross-traffic, bufferbloat, active queue management, policers and shapers, link layers with non-uniform packet forwarding behaviours, …)
- Designed network lab environments for controlled data networking experiments exploring large, multi-parameter test spaces (including real-time data capture & logging, automation and orchestration for unsupervised operation, …)
- Experience with FreeBSD or similar *NIX-like system configuration & administration (including kernel patching & rebuilding, installing and tuning networked user-space applications, tuning transport/TCP parameters to match network conditions, …)
- Ability to articulate for diverse audiences the technical and business implications of both production and lab-based data networking experiments (eg. industry white papers, keynotes, peer-reviewed industry/academic conference or journal papers, …)
- At minimum a Bachelor of Science (Computer Science), Bachelor of Engineering (Electronic or Telecommunications) or closely related Bachelor degree
It will help if you have:
- Intimate familiarity with common Transmission Control Protocol (TCP) implementations, and commonly deployed performance tuning heuristics
- Experience designing and executing defensible IP, transport and/or application-layer measurement studies of Internet characteristics
- Experience with system control, data manipulation and presentation tools (unix shell, Python, R, …) and applying statistical tools to the modelling of production traffic patterns
- Designed and implemented network lab trials that plausibly emulate real-world conditions (such as with TEACUP, https://sourceforge.net/projects/teacup/, FLENT, https://flent.org/, or similar)
- Experience with the IETF & IRTF processes for development of networking protocols
- Experiencing partnering with remote engineering or product teams across multiple time zones
- Masters or PhD (with very strong focus on practical, experimental data networking research)
Additional requirements:
- Limited travel