Robinson Mann

The Database Access Platform team builds and operates a flexible query gateway that facilitates data abstractions to operate at sub-millisecond latencies while allowing Netflix microservices to more easily store, consume, and manage their data. This team holds a substantial responsibility in enabling Netflix microservices to satisfy their ever-growing and evolving data needs. This team is passionate about distributed data systems technology. We are active in the open source community and believe in operating what we own. We are a small team responsible for business critical systems and are committed to a culture of feedback and engineering.

- Worked on AWS Key Management Service (KMS), a Tier-0 AWS Service with >10,000,000 TPS Globally, 99.999% Availability, and P99.9 latencies of 30 milliseconds, on the Data, Analytics, and Observability team. - Decreased data propagation times from 3 hours to 3 minutes (98%) in a 60,000GB/hour service log pipeline by spearheading the transition to a streaming architecture for metrics which were identified as critical. Great care was taken to not impact overall Latency, CPU, Memory, Network, and GC Heap Size Utilizations. Created cross-service log tracing by onboarding other teams to the same architecture and consolidating data in a shared data lake. - Developed a real-time Online Analytic Processing (OLAP) replica of the primary datastore inside of RDS PostgreSQL. This replica granted KMS leadership, business partners, and security engineers access to the primary datastore for the first time and reduced 1 hour queries to <1 minute. Saved >80 hours/month for KMS Engineers. - Built a fully managed Online Event Processing (OLEP) framework with guaranteed delivery to enable distributed asynchronous behavior for control plane APIs within KMS. This system had >99.99% availability and successfully reduced control plane latencies by offloading work from the critical paths of KMS APIs. Used for 7 different projects across 4 different teams. - Reduced P99 control plane latency by 75%, enabled doubling frontend capacity on the existing architecture, and reduced datastore writes by 50% by building event driven resource quota enforcement. - Lead Engineer in a 6-person team. Performed annual project planning, mentoring, hiring, task triage, operational tasks/oncall, design reviews, launch reviews, benchmarking, capacity planning, and deployments. Delivered presentations within the organization on software engineering best practices and the team’s software architecture. Coordinated across multiple teams and stakeholders to align business needs with application architecture.

- Worked with the Alexa Personalization Team to build personalized language models for individual customers based on their selected languages, installed apps, contact names, etc. - Modernized an analysis workflow from being single-machined to running in a distributed fashion using Hadoop MapReduce. Reduced the time of the workflow from 36 hours to 10 minutes. - Designed and implemented the infrastructure to compare different text relevancy algorithms on datasets. My work provided insight into how our production neural networks compared to traditional algorithms.

- Worked with the Alexa Personalization Team to improve the internal model benchmarking software for Alexa's ASR/NLU.

- Designed, implemented and deployed an API to modify product supply levels in real time on Amazon.com for the Fulfillment by Amazon (FBA) Team. This API has been used to protect customers from purchasing inventory we suspect may not exist. Created a web application to expose this API to internal customers.

- Created an API to retrieve detailed information about inbound shipments for the Fulfillment by Amazon (FBA) Team. This included designing the schema for the datastore. Created a web application to expose this API to internal customers.

Graduated with Distinction - Dean's Honours List. GPA: 88.48%. Math & CS GPA: 89.12%