Lead the technical strategy and implementation of solutions to reduce noise, minimize taint, improve determinism within the Search serving stack (QRS, QRewrite, MaRS, Muppet).
Drive the architectural evolution of the evaluation serving stack to support emerging multi-modal, multi-turn flow requirements, while collaborating with partner teams across Search Serving, Search Intelligence, Search Evals, Search AI Infra to diagnose and resolve issues affecting eval fidelity.
Establish best practices for determinism within serving components (e.g., Superroot, MaRS) to minimize product non-determinism in production-like evaluation environments.
Mentor and provide technical guidance to fellow engineers on the team, and contribute to the team's on-call rotation to support the stability of the eval environment.
Own and drive the reliability program for Gemini model release cycles, AIO/AIM launches, co-ordinating dependencies across infrastructure and evaluation teams to ensure high-stability and high-utility signals.
Minimum qualifications:
Bachelor’s degree or equivalent practical experience.
5 years of experience with software development in one or more programming languages.
3 years of experience testing, maintaining, or launching software products, and 1 year of experience with software design and architecture.
Experience with large-scale C++/Java serving systems and distributed system observability (e.g., Dapper traces, Monarch metrics).
Experience working on system-wide reliability or performance initiatives.
Experience designing infrastructure systems for various applications, and evaluating and selecting infrastructure components.
Preferred qualifications:
Master's degree or PhD in Computer Science or related technical field.
Experience as a technical leader with a track record of delivering results and making progress in an ambiguous problem space, and driving technical strategy and road map development.
Ability to demonstrate technical leadership, mentor and grow other engineers, navigate technical ambiguity and drive consensus among multiple partner teams with competing priorities.
