Aaron Lobb Email & Phone Number

New York, New York, United States

Databases Org

New York, New York, United States

(E5) Conveyor (Deployment Infrastructure)• Developed reliable back-end infrastructure components to provide frequent pushes and positive user experience for the deployment system for all back-end systems at Facebook.• Implemented push metrics for build commit age to support more accurate binary staleness tracking anddebugging of push pipeline process. These were used by multiple teams across Facebook to assess andmonitor push delays and overall push pipeline more efficiently.• Convinced the Conveyor team to prioritize building a long-term solution for bad package protection, whichwould lead to the dozens of teams currently disabling the protection mechanism to reenable and reducetheir vulnerability to known bugs.• Built conversion API to integrate pushes from existing deployment system with new declarativeinfrastructure, a Terraform-like internal system.• Leveraged experience and empathy as a former client of the Conveyor team to advocate for other internalConveyor clients and communicate their needs to the team.• Technologies: C++, Rust, Python, Hacklang, distributed systems, internal cloud

New York, New York, United States

(E4-E5) IGML (Sourcing and Ranking Engine)• Built full-stack components and tooling for a large scale, mission-critical Instagram service with hundredsof developers making changes to the code base.• Improved push process by creating push metrics and SLOs and improving company build infra to supportopt-in hourly builds, improving push frequency and reliability (~1 push every 3-4 days to >1 push per day).• Partnered with Conveyer and CI teams to incorporate IGML changes into underlying infrastructuresystems, allowing all infrastructure teams to benefit.• Led effort to split on-call into production and push support. Led weekly retrospectives focused on rootcause analysis, categorizing issues, and ensuring follow-ups were completed. Participated in bothproduction and push on-call schedules in addition to project work.• Rewrote developer tooling for testing queries onto modern and supported company tech stack, fixedprivacy issues, allowed custom entity rendering, and provided extensible integration for new query layers(e.g., query optimization and understanding).• Implemented caching for query filters resulting in 30% reduced downstream QPS for certain query types.• Brought reliability focused mindset to the team, influencing other engineers to prioritize reliability issues,perform proper RCA, and track resolution.• Technologies: C++, Hacklang, Javascript, React, Python, distributed systems, internal cloud

New York, New York, United States

(E3-E4) Distributed Tracing Infrastructure• Led storage migrations of tracing data (5 PB/month).• Improved trace data usability by migrating trace processing from custom DSL to client-written Python.• Rewrote trace filtering to facilitate migration to smaller machines (256 GB memory to 32 GB). Additionaloptimizations reduced machine requirements from three to one.• Implemented optimizations for read path resulting in 30-40% reduced latency improvement for tracevisualizations.• Technologies: Java, Python, Hacklang, C++, distributed systems, internal cloud

Menlo Park, California, United States

• Added Java API for querying distributed KV storage.• Rewrote parser and AST components, unifying multiple legacy formats and allowing future work for query optimization.

New York

• Integrated C++ legacy bond pricing system with updated systems.

San Mateo, California, United States